Lower Risk of Heart Failure and Death in Patients Initiated on SGLT-2 Inhibitors Versus Other Glucose-Lowering Drugs: The CVD-REAL Study

Transcription

1 /CIRCULATIONAHA Lower Risk of Heart Failure and Death in Patients Initiated on SGLT-2 Inhibitors Versus Other Glucose-Lowering Drugs: The CVD-REAL Study Running Title: Kosiborod et al.; CVD-REAL Study Downloaded from by guest on July 5, 2017 Mikhail Kosiborod, MD 1 ; Matthew A. Cavender, MD, MPH 2 ; Alex Z. Fu, PhD 3 ; John P. Wilding, MD, PhD 4 ; Kamlesh Khunti, MD, PhD 5 ; Reinhard W. Holl, MD, PhD 6 ; Anna Norhammar, MD 7 ; Kåre I. Birkeland, MD, PhD 8,9 ; Marit Eika Jørgensen, MD, PhD 10,11 ; Marcus Thuresson, PhD 12 ; Niki Arya, MSc 13 ; Johan Bodegård, MD, PhD 14 ; Niklas Hammar, PhD 7,15 ; Peter Fenici, MD PhD 16 on behalf of the CVD-REAL Investigators and Study Group 1 Saint Luke's Mid America Heart Institute and University of Missouri-Kansas City, Kansas City, MO; 2 University of North Carolina, Chapel Hill, NC; 3 Georgetown University Medical Center, Washington DC; 4 University of Liverpool, Liverpool, United Kingdom; 5 University of Leicester, Leicester, United Kingdom; 6 University of Ulm, Ulm, Germany; 7 Karolinska Institutet, Stockholm, Sweden; 8 University of Oslo, Oslo, Norway; 9 Oslo University i Hospital, Oslo, Norway; 10 Steno Diabetes es Center, Copenhagen, n, Gentofte, e Denmark; 11 National al Institute of Public Health, Southern Denmark University, Denmark; 12 Statisticon AB, Uppsala, Sweden; 13 AstraZeneca, Gaithersburg, MD; 14 AstraZeneca, Oslo, Norway; 15 AstraZeneca Gothenburg, Sweden; 16 AstraZeneca, Cambridge, United Kingdom Address for Correspondence: Mikhail Kosiborod, MD Saint Luke s Mid America Heart Institute 4401 Wornall Road Kansas City MO, Tel: Fax: mkosiborod@saint-lukes.org 1

2 /CIRCULATIONAHA Abstract Downloaded from by guest on July 5, 2017 Background Reduction in cardiovascular death and hospitalization for heart failure (HHF) was recently reported with the sodium-glucose co-transporter-2 inhibitor (SGLT-2i) empagliflozin in type 2 diabetes patients with atherosclerotic cardiovascular disease. We compared HHF and death in patients newly initiated on any SGLT-2i versus other glucose lowering drugs (oglds) in six countries to determine if these benefits are seen in real-world practice, and across SGLT-2i class. Methods Data were collected via medical claims, primary care/hospital records and national registries from the US, Norway, Denmark, Sweden, Germany and the UK. Propensity score for SGLT-2i initiation was used to match treatment groups. Hazard ratios (HRs) for HHF, death and their combination were estimated by country and pooled to determine weighted effect size. Death data were not available for Germany. Results After propensity matching, there were 309,056 patients newly initiated on either SGLT-2i or ogld (154,528 patients in each treatment group). Canagliflozin, dapagliflozin, and empagliflozin accounted for 53%, 42% and 5% of the total exposure time in the SGLT-2i class, respectively. Baseline characteristics were balanced between the two groups. There were 961 HHF cases during 190,164 person-years follow up (incidence rate [IR] 0.51/100 person-years). Of 215,622 patients in the US, Norway, Denmark, Sweden, and UK, death occurred red in (IR 0.87/100 person-years), and HHF or death in 1983 (IR 1.38/100 person-years). Use of SGLT-2i, versus oglds, was associated with lower rates of HHF (HR 0.61; 95% CI ; p<0.001); death (HR 0.49; 95% CI ; p<0.001); and HHF or death (HR 0.54; 95% CI , p<0.001) with no significant heterogeneity by country. Conclusions In ons In this large multinational study, treatment with SGLT-2i versus oglds was associated ated with a lower risk of HHF and death, suggesting stin that the benefits e seen en with empagliflozin in a randomized trial may be a class effect applicable plicable to a broad population pulation of T2D patients in real-world rld practice c (NCT ). Clinical ical Trial Registration URL: U ClinicalTrials.gov; icaltr Unique Identifier: ier: NCT Key Words: death; diabetes mellitus; heart failure; observational studies; SGLT2 inhibitor 2

3 /CIRCULATIONAHA Clinical Perspective Downloaded from by guest on July 5, 2017 What is new? This is the first large real world study of more than 300,000 patients with Type 2 diabetes (T2D), both with and without established cardiovascular disease, from routine clinical practice across six countries - evaluating the outcomes of hospitalization for heart failure (HHF) and all-cause death in patients with T2D treated with SGLT-2 inhibitors (SGLT- 2i) versus other glucose-lowering drugs (ogld) The distribution of exposure time for the various SGLT-2i compounds (for HHF outcome) was 53% for canagliflozin, 42% for dapagliflozin, and ~5% for empagliflozin, with substantial inter-country variability. What are the clinical implications? Treatment with SGLT-2i versus ogld was associated with a 39% relative risk reduction in HHF, a 51% reduction in all-cause death, and a 46% reduction in the HHF or death composite, os consistent with the effects previously reported in a randomized clinical trial of empagliflozin Approximately 87% of patients did not have known cardiovascular cular disease, e suggesting possible cardiovascular benefits for a broad population of T2D patients The lower rates of HHF and death associated with SGLT-2i treatment t t appear likely l classrelated as there was no significant heterogeneity across countries, despite geographic variations in the use of specific SGLT-2i (~76% canagliflozin in US and ~92% dapagliflozin in Europe. 3

4 /CIRCULATIONAHA Type 2 diabetes (T2D) remains a major risk factor for cardiovascular disease (CVD) 1, 2, and overall mortality, 3, 4 despite advances in treatment. 5-7 Heart failure is an especially common complication of T2D, 8-10 with particularly poor outcomes and five-year survival rates of <25%. 11 This highlights the need for novel treatments that not only improve glycemic control, but also reduce the risk of CVD, including heart failure. Although higher HbA1c is associated with greater risk of CVD, 12 intensive glucose control has failed to reduce the development of heart failure, and cardiovascular-related or allcause death. However, EMPA-REG OUTCOME, a prospective randomized controlled trial Downloaded from by guest on July 5, 2017 (RCT) in patients with T2D and established atherosclerotic CVD, demonstrated a substantial reduction in CV death and hospitalization for heart failure (HHF) with the sodium-glucose ucose cotransporter-2 inhibitor (SGLT-2i), empagliflozin, 13 within a short follow-up period. The mechanisms of these benefits, while unclear, were almost certainly not due to glucose lowering, given a very small difference fe e in HbA1c levels between empagliflozin- lozi and placebo-treated eate ted patients and early separation of the event curves. Following the EMPA-REG OUTCOMES OMES trial, several critical questions remain, with substantial clinical implications. First, the applicability of findings to real-world clinical practice (where patients receive standard of care with various other glucose-lowering drugs [oglds]) is unclear. Second, it is unknown whether the observed benefits are specific to empagliflozin, or represent a class effect. Finally, since EMPA-REG OUTCOME only included patients with established CVD, it remains to be seen if similar benefits can be expected in T2D patients with a broader cardiovascular risk profile. Using data from multiple countries in The Comparative Effectiveness of Cardiovascular Outcomes (CVD-REAL) study (NCT ) we compared the risk for HHF, death, and the 4

5 /CIRCULATIONAHA combined endpoint of HHF or death in patients with T2D who were new users of SGLT-2i versus oglds in real-world practice. Methods Data Sources De-identified health records across six countries (US, Germany, Sweden, Norway, Denmark and UK) were analyzed. In the US, Truven Health MarketScan Claims and Encounters and linked Medicare Supplemental and Coordination of Benefits databases were used, which included Downloaded from by guest on July 5, 2017 enrollment and demographic information, inpatient and outpatient medical, and outpatient pharmacy claims from >300 large self-insured US employers and >25 US health plans. In Germany, the Diabetes Prospective Follow-Up (DPV) initiative is a quality assessment registry in individuals with diabetes and employs standardized documentation and objective comparison of quality indicators, icator with 452 centers e participating. ating. In Sweden, en Norway and Denmark mandatory national al full-population pulation registries of each respective country were e used, with linked Prescribed Drug Registers covering all drugs dispensed, National Patient Registers covering all hospitalizations and specialized outpatient care, and Cause of Death Registers In the UK, records from CPRD and THIN datasets were used, which included primary care data from >670 general practices linked with hospitalization and mortality registries. Additional details of the individual datasets can be found in Supplemental Appendices. Patient Cohort Patients with T2D (diagnosis codes in Supplemental Tables 1 2) that were newly started on either SGLT-2i or newly started on oglds were selected from each dataset beginning on the date of first prescription or pharmacy dispensation of an SGLT-2i or a new ogld in each of the 5

6 /CIRCULATIONAHA countries (start date ranged from November 2012 in the UK to July 2013 in Sweden). New users were defined as individuals prescribed/filling a prescription (as initial or add-on therapy) for any SGLT-2i (canagliflozin, dapagliflozin, or empagliflozin) or ogld (any other oral or injectable medication), including fixed-dose combinations, with no issued prescriptions of that medicine class during the preceding year (in Germany, with no prior documentation in the medical record of using that medicine class within previous six months). Additional inclusion criteria were age ription date for new SGLT-2i or new ogld), and >1 year data history in the database prior to index date. Patients with Type 1 or gestational Downloaded from by guest on July 5, 2017 diabetes were excluded. Patients were followed from the index date until end of the index treatment (for the on-treatment analysis), migration/leaving the practice/database, last date of data collection, outcome date, or censoring date (range from September 2015 in the US to November 2016 in Sweden). Outcomes Primary outcome was HHF assessed sessed in all countries. In the US, UK and Germany, HHF was defined as hospital admissions for heart failure (defined using primary discharge diagnosis codes in the US, 18 primary discharge diagnosis codes and documentation from the electronic health records in the UK, as defined in the Supplemental Appendix, and documentation in the electronic health records in Germany). In the Nordic countries (Sweden, Norway, Denmark), hospitalization for heart failure was defined by any hospital visit, in- or outpatient (i.e. prognostically equivalent outpatient HF event ), 19 with a registered primary diagnosis of heart failure (defined using diagnosis codes for heart failure events as detailed in the Supplemental Appendix, and validated independently in all three countries) Secondary outcomes included all-cause death; and composite of HHF or all-cause death (time-to-first-event), evaluated in all 6

7 /CIRCULATIONAHA countries, except Germany. In the US, all-cause death was identified using the MarketScan Mortality File in which information from the Social Security Administration is integrated with the insurance enrollment and claims data, supplemented by claims for in-hospital deaths, covering ~61% of the overall US-based propensity-matched patient cohort. Characteristics of US patients with and without vital status were similar (Supplemental Table 3), indicating data missing completely at random due to administrative reasons. Statistical Analysis Baseline characteristics of patients in the SGLT-2i and ogld groups were analyzed using Downloaded from by guest on July 5, 2017 descriptive statistics. Categorical variables were described by frequencies and percentages, and continuous variables using mean (± SD). For continuous variables such as age, the overall mean across all databases was a summary estimate of country-specific means, weighted according to the number of patients in each respective database. For the SGLT-2i group, the percentage e of individual iv idua agents and their respective contributions tion ons to the overall SGLT2i exposure time; and for the ogld group, the percentage of individual dual drug classes; were summarized by country/geographic region and overall. A non-parsimonious propensity score was developed (separately within each country) for being initiated on an SGLT-2i, to minimize confounding. Variables that may have affected treatment assignment or outcomes were included in the propensity score (Supplemental Table 4). 23 Based on propensity scores, patients receiving SGLT-2i were matched 1:1 with those receiving oglds. Nearest neighbor caliper width of 0.25 multiplied by the SD of the propensity score distribution was used for the matching. 23 In Sweden, Norway and Denmark an automated balance optimization method using the function Match (in package Matching) in R and a caliper of 0.2 were used for matching. The adequacy of propensity matching was assessed by 7

8 /CIRCULATIONAHA standardized differences of post-match patient characteristics. A significant imbalance was considered to be present if a >10% standardized difference was present between the two groups after propensity match. 24 Incidence analyses of HHF, death and composite of HHF or death were conducted by treatment group. Only the first episode of each outcome was included, and the crude incidence rate (IR) in each group was calculated as the number of incident events divided by the total number of person-years at risk, and expressed per 100 person-years with 95% CI. Time to first event for the SGLT-2i and ogld groups were compared using Cox proportional hazards models Downloaded from by guest on July 5, 2017 and presented as hazard ratios (HR) and 95% CI for each outcome separately within each country. The primary analysis used on-treatment approach where patients were followed from the start of an index treatment and censored at the end of that treatment plus a grace period (duration of last issued prescription). r ption) The HRs (95% CI) for each of the endpoints nts from each individual dual country were then pooled together for an overall weighted summary, 25 in which random-effects models with inverse variance weighting for each country were implemented. 26 Forest plots displaying the country-specific HRs (95% CI) along with the pooled overall HR (95% CI) were produced. Multiple sensitivity analyses were conducted: first, the HR (95% CI) within each country, and for each outcome, were examined after adjusting the crude propensity-matched estimates for multiple covariates that may have confounded the relationship between treatment and outcome. The adjusted HR (95%CI) from each country were pooled and meta-analyzed using the same method as described above. Second, the analyses for each outcome were repeated using Intent- To-Treat (ITT)-analysis, in which patients were followed after discontinuation of index 8

9 /CIRCULATIONAHA treatment Third, the analyses for HHF were repeated after stepwise removal of specific ogld classes from the comparator group, to examine whether a specific ogld class contributed disproportionately to the results. Stepwise elimination was performed in the following sequence: thiazolidinediones, thiazolidinediones+insulin, thiazolidinediones+insulin+sulphonylureas. Fourth, HHF analyses were repeated after excluding patients treated with GLP-1RA at baseline from SGLT-2i and ogld groups. Fifth, primary analyses were repeated separately in the US and Europe. Finally, the association between treatment with SGLT-2i and ogld was reexamined separately for patients that had both in- and outpatient hospital visits for HF, and those that had Downloaded from by guest on July 5, 2017 only inpatient hospital visits for HF in Sweden (as these could not be separated in Norway and Denmark; and only inpatient HF visits were analyzed in other countries). For power calculations, see Supplemental Appendix. Due to the de-identified nature of patient records, informed consent was not obtained. Analyses of de-identified data were conducted in accordance with local laws and regulations, and received approvals prov als from respective Scientific/Ethics/Data ific/ethic ics/ Protection Committees. Country-specific ic analyses were conducted cted by independent academic/statistical groups. The meta-analyses were conducted by Statisticon, i and validated by the independent academic statisticians at Saint Luke s Mid America Heart Institute. Results Study population A total of 1,392,254 new SGLT-2i or ogld users were identified; 166,033 SGLT-2i, and 1,226,221 ogld overall and by country (Figure 1 and Supplemental Figure 1). Before propensity match (Supplemental Table 5) patients initiated on SGLT-2i were younger, less likely to have chronic kidney disease or cardiovascular complications, but more likely to have 9

10 /CIRCULATIONAHA microvascular disease. Greater proportions of patients initiated on SGLT-2i vs. ogld received statins and anti-hypertensive drugs, and lower proportions received loop diuretics. Patients on SGLT-2i were more likely to be treated with other glucose-lowering medication classes at baseline. The overlap in propensity scores between groups before and after the propensity match is shown in Supplemental Figures 2 3. Baseline characteristics were well-balanced between groups post-matching overall and by country (Table 1 and Supplemental Table 6), with standardized differences for most variables being <10% (Supplemental Figure 4). Pre- and post-match standardized differences are shown in Downloaded from by guest on July 5, 2017 Supplemental Table 7. Mean age was 57 years, 44% were women, and 13% had established CVD. Overall, 67% of patients received statins, 80% anti-hypertensive medications, 74% with ACEi/ARBs, and 79% metformin. The composition of SGLT-2i agents is shown in Supplemental Table 8 and the composition of the index medications ions in the ogld group in Supplemental Tables The composition of SGLT-2i agents in terms of total exposure time was balanced between canagliflozin and dapagliflozin, with <7% total exposure attributable tabl to empagliflozin for all outcomes (Figure 2). SGLT-2i and HHF A total of 309,056 patients (154,528 in each group) were identified after propensity matching. Canagliflozin, dapagliflozin, and empagliflozin accounted for 53%, 42% and 5% of the total exposure time in the SGLT-2i class, respectively (Figures 2A C). Over 190,164 person-years follow-up, there were 961 HHF events (IR 0.51/100 personyears; Supplemental Table 11; IR by treatment group in Supplemental Table 12). Mean duration of follow up for HHF was 239 days in the SGLT-2i group and 211 days in the ogld group 10

11 /CIRCULATIONAHA Downloaded from by guest on July 5, 2017 (Supplemental Table 13). Initiation of SGLT-2i vs. ogld was associated with a lower risk of HHF (pooled HR 0.61, 95%CI ; P<0.001; Figure 3A). HRs favored SGLT-2i in each country (P-value for heterogeneity 0.17). SGLT-2i and all-cause death A total of 215,622 patients (107,811 in each group) were identified. Canagliflozin, dapagliflozin, and empagliflozin accounted for 42%, 51% and 7% of SGLT-2i exposure time, respectively (Figures 2A C). Over 153,990 person-years of follow up, there were 1334 events (IR 0.87/100 personyears; Supplemental Table 11; IR by treatment group in Supplemental Table 12). Mean duration of follow up was 271 days in the SGLT-2i group and 251 days in the ogld group (Supplemental Table 13). Initiation of SGLT-2i vs. ogld was associated with a lower risk of death (pooled HR 0.49, 95%CI , P<0.001; Figure 4A). HRs favored SGLT-2i in each country (P-value for heterogeneity erog enei eity 0.09). SGLT-2i and composite outcome ome of HHF or death For the composite outcome, the number of patients was identical to the all-cause death analysis. Canagliflozin, dapagliflozin, and empagliflozin accounted for 45%, 49% and 6% of SGLT-2i exposure time, respectively (Figures 2A C). Over 143,342 person-years of follow up, there were 1983 events (IR 1.38/100 personyears; Supplemental Table 11; IR by treatment group in Supplemental Table 12). Mean duration of follow up was 253 days in the SGLT-2i group and from 233 days in the ogld group (Supplemental Table 13). Initiation of SGLT-2i vs. ogld was associated with a lower risk of HHF or death (pooled HR 0.54, 95%CI , P<0.001; Figure 4D). HRs favored SGLT-2i in each country (P-value for heterogeneity 0.17). 11

12 /CIRCULATIONAHA Sensitivity analyses For all three outcomes, similar results were found after multivariate adjustment (Figures 3B, 4B, 4E), using an ITT-approach (Figures 3C, 4C, 4F) and stepwise removal of specific ogld classes (Supplemental Figures 5 6). Comparisons within geographic regions yielded similar results (Supplemental Figure 7). The association between treatment with SGLT-2i vs. ogld and lower risk of HHF was consistent among patients that had both in- and outpatient hospital visits for HF, and those who had only inpatient hospital visits for HF in Sweden (Supplemental Table 14). Downloaded from by guest on July 5, 2017 Discussion In this large contemporary analysis of real-world clinical practice across six countries, within in a well-matched sample of over 300,000 T2D patients and nearly 200,000 patient-years of observation, initiation of SGLT-2i versus oglds was associated with a 39% lower incidence of HHF. Since the overwhelming majority of patients ts did not have established CVD, this suggests that the benefits of SGLT-2i on the prevention of heart failure may extend to lower risk patients than those enrolled in randomized trials so far. These findings were unchanged nged after additional multivariable adjustment, and in multiple sensitivity analyses. Specifically, the results were unchanged after sequential removal of several ogld classes from the comparator group suggesting that the differential outcomes observed are unlikely to reflect adverse effects of comparator drugs, but are rather associated with benefit from SGLT-2i. Furthermore, results were consistent across countries, regardless of variability in healthcare systems and use of specific SGLT-2i (predominantly canagliflozin in the US; dapagliflozin in Europe), suggesting an association with the class rather than any single agent. Importantly, initiation of SGLT-2i 12

13 /CIRCULATIONAHA versus oglds was also associated with a 51% lower rate of all-cause death, and 46% lower rate of the combined endpoint of HHF or all-cause death. Although intensive glucose lowering has, in randomized trials, failed to reduce what are, arguably, some of the most important outcomes in patients with T2D (all-cause death and incident heart failure), results from the EMPA-REG OUTCOME trial demonstrated that such benefits are achievable within a short time frame with an SGLT-2i, likely via non-glycemic mechanisms. Ultimately, the main goals of treating patients with T2D are to prolong life, and improve quality of life. Given that CVD (including heart failure) is a leading cause of Downloaded from by guest on July 5, 2017 mortality/morbidity in T2D, the results of the recent cardiovascular outcomes trials (CVOTs) suggest that the time has come to shift from the narrow focus on HbA1c to a more comprehensive focus in which treatments proven to improve important outcomes (especially mortality) are prioritized. Our findings address dres s several key unanswered questions with regards rds to the potential tial role of SGLT-2i in the management e of T2D, with important clinical ic implications. First, our results demonstrate that the effects associated with the use of SGLT-2i in regards to HHF and all-cause death are remarkably similar in real-world practice to those seen in the EMPA-REG OUTCOME trial. Second, we found no significant heterogeneity in results across countries, despite geographic variations in the use of specific SGLT-2i, suggesting that the associated lower risks for cardiovascular outcomes are likely class-related. Indeed, for all outcomes evaluated, empagliflozin contributed <7% of total exposure time. Third, we evaluated a broader cardiovascular risk population in general practice, where the overwhelming majority (87%) had no established CVD, suggesting that lower risk patients may derive similar benefits with SGLT- 2i, as those with higher risk. If confirmed by data from ongoing trials (CANVAS 30, 13

14 /CIRCULATIONAHA NCT ; DECLARE, NCT ; VERTIS, NCT ), this would have substantial impact on clinical practice. In this regard, we see the data produced from carefully conducted, methodologically rigorous, large multi-country epidemiologic studies, as complementary to those generated by clinical trials, as they help establish the real-world effectiveness of treatments in a broad population of patients from clinical practice. Indeed, the importance of such studies for a broad range of objectives, including evaluation of treatment effects and outcomes, and their potential for complementing the knowledge gained from clinical trials is being increasingly recognized, 31 and the terminology describing these as real-world Downloaded from by guest on July 5, 2017 evidence, has recently been accepted by major international regulatory bodies To our knowledge, CVD-REAL is the first large study addressing the real world effectiveness (rather than efficacy) of SGLT-2i on specific outcomes of HHF and all-cause death across multiple countries. Given that SGLT-2i is a novel class, real-world experience is limited. Single-country data were previously reported rted with a specific ic SGLT-2i (dapagliflozin) agli in) in Sweden; however that study was limited by the smaller number of patients, and focused on different ferent outcomes (hypoglycemia and composite of CVD). 35 The size of our sample and ability to pool data from diverse sources, allowed us to collect a large number of events, and examine the stability of results across various cardiovascular outcomes, multiple countries (with variable use of specific SGLT-2i), and perform numerous sensitivity analyses. Our findings should be examined within the context of several potential limitations. First, given the observational nature of the study, and despite robust propensity-matching and multiple sensitivity analyses, a possibility of residual, unmeasured confounding, cannot be excluded. Second, we focused on HHF and all-cause death, and did not examine other events, such as MI and stroke. However, heart failure is arguably the most lethal T2D complication, 9, 36 and 14

15 /CIRCULATIONAHA associated with particularly poor survival. 11 Third, we did not examine safety. Fourth, despite a large number of patient-years of follow up, SGLT-2i experience in real-world practice is still relatively limited; longer-term follow up will be required to examine if effects are sustained over time. Fifth, there were differences in the definitions of HHF across countries; however the results were consistent across countries, and in sensitivity analyses specifically performed to examine these differences. Finally, our study did not address the mechanisms linking use of SGLT-2i and associated cardiovascular benefits. However, this knowledge gap is being examined by mechanistic investigations across the class Downloaded from by guest on July 5, 2017 Conclusion In this large multinational study, treatment with SGLT-2i versus oglds was associated with lower rates of HHF and death, suggesting that the benefits previously reported with empagliflozin in the context of a randomized trial may be applicable p able to a broad population p on of patients with T2D in real-world practice. ce. The lack of heterogeneity in results across countries, despite geographic variations in the use of specific ic SGLT-2i, suggests a class effect f for SGLT2i. Acknowledgements The authors wish to acknowledge Betina T. Blak, Sara E. Dempster, Markus F. Scheerer, Karolina Andersson-Sundell, Kelly Bell, Eric T. Wittbrodt, Luis Alberto García Rodríguez, Lucia Cea Soriano, Oscar Fernándex Cantero, Ellen Riehle, Brian Murphy, MS Esther Bollow, Hanne Løvdal Gulseth, Bendix Carstensen, Fengming Tang, Kevin Kennedy and Sheryl L Windsor for their tireless contribution to the country level analyses, quality check validation and results interpretation. Data validation was independently conducted by MAHI, an external 15

16 /CIRCULATIONAHA academic institution. All authors had access to the full data package and are responsible for data interpretation and conclusions. Additional details on databases and sensitivity analyses are provided in the Supplementary materials. Editorial support was provided by Róisín O Connor and Mark Davies, inscience Communications, Springer Healthcare, and funded by AstraZeneca Sources of Funding This work was supported by AstraZeneca. Downloaded from by guest on July 5, 2017 Disclosures MK, advisory boards for AstraZeneca, Boehringer Ingelheim, Sanofi, Glytec, Novo Nordisk, ZS Pharma, GSK, Amgen, Eisai and Merck (Diabetes); consultant for AstraZeneca, Sanofi, and ZS Pharma, research grants for AstraZeneca and Boehringer Ingelheim. MAC, personal fees from Merck and AstraZeneca. eca. AZF, grants from AstraZeneca and Merck; personal fees es from Asclepius Analytics and Complete HEOR Services. es. JPW, lecture fees from Astellas, AstraZeneca, Boehringer Ingelheim, Janssen, Lilly, Novo Nordisk, Orexigen, Sanofi; and consultancy (Institutional) from AstraZeneca, Boehringer Ingelheim, Janssen, Lilly, and Orexigen; grants to institution from Takeda, Novo Nordisk and AstraZeneca. KK, consultant and speaker for Astra Zeneca, Novartis, Novo Nordisk, Sanofi-Aventis, Lilly, Merck Sharp & Dohme, Janssen and Boehringer Ingelheim; grants in support of investigator and investigator initiated trials from Astra Zeneca, Novartis, Novo Nordisk, Sanofi-Aventis, Lilly, Boehringer Ingelheim, Merck Sharp & Dohme and Roche; advisory boards for Astra Zeneca, Novartis, Novo Nordisk, Sanofi-Aventis, Lilly, Merck Sharp & Dohme, Janssen and Boehringer Ingelheim. RWH, grants from AstraZeneca. AN, personal fees from AstraZeneca for this study; honorarium 16

17 /CIRCULATIONAHA for lectures and advisory board meetings for Novo Nordisk, Boehringer Ingelheim, and Lilly. KIB, grants to his institution from AstraZeneca for this study. MEJ, shareholder of Novo Nordisk, was employed by Steno Diabetes Center A/S until December , a research hospital working in the Danish National Health Service and owned by Novo Nordisk A/S; grants from Astra Zeneca and for lectures and consulting from Novo Nordisk, Sanofi, Lilly, Boehringer Ingelheim and Merck Sharp & Dohme. MT, employee of Statisticon who were under contract to AstraZeneca for this study. NA, JB, NH and PF, employees of AstraZeneca. Downloaded from by guest on July 5, 2017 References 1. Emerging Risk Factors C, Sarwar N, Gao P, Seshasai SR, Gobin R, Kaptoge S, Di Angelantonio E, Ingelsson E, Lawlor DA, Selvin E, Stampfer M, Stehouwer CD, Lewington S, Pennells L, Thompson A, Sattar N, White IR, Ray KK, Danesh J. Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: A collaborative meta-analysis of 102 prospective studies. Lancet. 2010;375: doi: /s (10) Schocken DD, Benjamin EJ, Fonarow GC, Krumholz HM, Levy D, Mensah GA, Narula J, Shor ES, Young JB, Hong Y, American Heart Association Council on E, Prevention, n American Heart Association Council on Clinical ical C, American Heart Association Council on Cardiovascular N, American Heart Association sociat io Council on High Blood Pressure R, Quality of C, Outcomes Research Interdisciplinary i Working G, Functional G, Translational Biology Interdisciplinary Working G. Prevention of heart failure: A scientific statement from the american heart association councils on epidemiology and prevention, clinical cardiology, cardiovascular nursing, and high blood pressure research; quality of care and outcomes research interdisciplinary working group; and functional genomics and translational biology interdisciplinary working group. Circulation. 2008;117: doi: /circulationaha Emerging Risk Factors C, Di Angelantonio E, Kaptoge S, Wormser D, Willeit P, Butterworth AS, Bansal N, O'Keeffe LM, Gao P, Wood AM, Burgess S, Freitag DF, Pennells L, Peters SA, Hart CL, Haheim LL, Gillum RF, Nordestgaard BG, Psaty BM, Yeap BB, Knuiman MW, Nietert PJ, Kauhanen J, Salonen JT, Kuller LH, Simons LA, van der Schouw YT, Barrett-Connor E, Selmer R, Crespo CJ, Rodriguez B, Verschuren WM, Salomaa V, Svardsudd K, van der Harst P, Bjorkelund C, Wilhelmsen L, Wallace RB, Brenner H, Amouyel P, Barr EL, Iso H, Onat A, Trevisan M, D'Agostino RB, Sr., Cooper C, Kavousi M, Welin L, Roussel R, Hu FB, Sato S, Davidson KW, Howard BV, Leening MJ, Rosengren A, Dorr M, Deeg DJ, Kiechl S, Stehouwer CD, Nissinen A, Giampaoli S, Donfrancesco C, Kromhout D, Price JF, Peters A, Meade TW, Casiglia E, Lawlor DA, Gallacher J, Nagel D, Franco OH, Assmann G, Dagenais GR, Jukema JW, 17

18 /CIRCULATIONAHA Downloaded from by guest on July 5, 2017 Sundstrom J, Woodward M, Brunner EJ, Khaw KT, Wareham NJ, Whitsel EA, Njolstad I, Hedblad B, Wassertheil-Smoller S, Engstrom G, Rosamond WD, Selvin E, Sattar N, Thompson SG, Danesh J. Association of cardiometabolic multimorbidity with mortality. JAMA. 2015;314: doi: /jama Kannel WB, Hjortland M, Castelli WP. Role of diabetes in congestive heart failure: The framingham study. Am J Cardiol. 1974;34: IDF Diabetes Atlas Group. Update of mortality attributable to diabetes for the idf diabetes atlas: Estimates for the year Diabetes Res Clin Pract. 2013;100: doi: /j.diabres NCD Risk Factor Collaboration (NCD-RisC). Worldwide trends in diabetes since 1980: A pooled analysis of 751 population-based studies with 4.4 million participants. Lancet. 2016;387: doi: /s (16) Tancredi M, Rosengren A, Svensson AM, Kosiborod M, Pivodic A, Gudbjornsdottir S, Wedel H, Clements M, Dahlqvist S, Lind M. Excess mortality among persons with type 2 diabetes. New Engl J Med. 2015;373: doi: /nejmoa Cavender MA, Steg PG, Smith SC, Jr., Eagle K, Ohman EM, Goto S, Kuder J, Im K, Wilson PW, Bhatt DL. Impact of diabetes mellitus on hospitalization for heart failure, cardiovascular events, and death: Outcomes at 4 years from the reduction of atherothrombosis for continued health (reach) registry. Circulation. 2015;132: doi: /circulationaha Fitchett D, Zinman B, Wanner C, Lachin JM, Hantel S, Salsali A, Johansen OE, Woerle HJ, Broedl UC, Inzucchi SE. Heart failure outcomes with empagliflozin in patients with type 2 diabetes at high cardiovascular risk: Results of the empa-reg outcome trial. Eur Heart J. 2016;37: doi: /eurheartj/ehv Green JB, Bethel el MA, Armstrong rong PW, Buse JB, Engel SS, Garg J, Josse R, Kaufman an KD, Koglin J, Korn S, Lachin JM, McGuire DK, Pencina MJ, Standl E, Stein PP, P Suryawanshi a S, Van de Werf F, Peterson ED, Holman RR. Effect of sitagliptin iptin on cardiovascular outcomes in type 2 diabetes. New Engl J Med ;373: doi: /NEJMoa Bertoni AG, Hundley WG, Massing MW, Bonds DE, Burke GL, Goff DC, Jr. Heart failure prevalence, incidence, and mortality in the elderly with diabetes. Diabetes Care. 2004;27: Matsushita K, Blecker S, Pazin-Filho A, Bertoni A, Chang PP, Coresh J, Selvin E. The association of hemoglobin a1c with incident heart failure among people without diabetes: The atherosclerosis risk in communities study. Diabetes. 2010;59: doi: /db Zinman B, Wanner C, Lachin JM, Fitchett D, Bluhmki E, Hantel S, Mattheus M, Devins T, Johansen OE, Woerle HJ, Broedl UC, Inzucchi SE, Investigators E-RO. Empagliflozin, cardiovascular outcomes, and mortality in type 2 diabetes. New Engl J Med. 2015;373: doi: /nejmoa Haheim LL, Helgeland J. Agreement between referral information and discharge diagnoses according to norwegian elective treatment guidelines - a cross-sectional study. BMC Health Serv Res. 2014;14:493. doi: /s Ludvigsson JF, Andersson E, Ekbom A, Feychting M, Kim JL, Reuterwall C, Heurgren M, Olausson PO. External review and validation of the swedish national inpatient register. BMC Public Health. 2011;11:450. doi: /

19 /CIRCULATIONAHA Downloaded from by guest on July 5, Norhammar A, Bodegard J, Nystrom T, Thuresson M, Eriksson JW, Nathanson D. Incidence, prevalence and mortality of type 2 diabetes requiring glucose-lowering treatment, and associated risks of cardiovascular complications: A nationwide study in sweden, Diabetologia. 2016;59: doi: /s y 17. Sundboll J, Adelborg K, Munch T, Froslev T, Sorensen HT, Botker HE, Schmidt M. Positive predictive value of cardiovascular diagnoses in the danish national patient registry: A validation study. BMJ Open. 2016;6:e doi: /bmjopen Lee DS, Donovan L, Austin PC, Gong Y, Liu PP, Rouleau JL, Tu JV. Comparison of coding of heart failure and comorbidities in administrative and clinical data for use in outcomes research. Med Care. 2005;43: Okumura N, Jhund PS, Gong J, Lefkowitz MP, Rizkala AR, Rouleau JL, Shi VC, Swedberg K, Zile MR, Solomon SD, Packer M, McMurray JJ. Importance of clinical worsening of heart failure treated in the outpatient setting: Evidence from the prospective comparison of arni with acei to determine impact on global mortality and morbidity in heart failure trial (paradigm-hf). Circulation. 2016;133: doi: /circulationaha Brynildsen J, Hoiseth AD, Nygard S, Hagve TA, Christensen G, Omland T, Rosjo H. [diagnostic accuracy for heart failure - data from the akershus cardiac examination 2 study]. Tidsskr Nor Laegeforen. 2015;135: doi: /tidsskr Kumler T, Gislason GH, Kirk V, Bay M, Nielsen OW, Kober L, Torp-Pedersen C. Accuracy of a heart failure diagnosis in administrative registers. Eur J Heart Fail. 2008;10: doi: /j.ejheart Ingelsson son E, Arnlov J, Sundstrom J, Lind L. The validity of a diagnosis of heart failure in a hospital discharge register. r. Eur J Heart Fail. 2005;7: ;7 doi: /j.ejheart Stuart EA. Matching methods for causal al inference: ence A review and a look forward. Stat Sci. 2010;25:1-21. doi: /09-sts Normand ST, Landrum MB, Guadagnoli E, Ayanian JZ, Ryan TJ, Cleary PD, McNeil BJ. Validating recommendations for coronary angiography following acute myocardial infarction in the elderly: A matched analysis using propensity scores. J Clin Epidemiol. 2001;54: DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled clinical trials. 1986;7: Viechtbauer W. Conducting meta-analyses in r with the metafor package. J Stat Softw. 2010;36: Hernan MA, Alonso A, Logan R, Grodstein F, Michels KB, Willett WC, Manson JE, Robins JM. Observational studies analyzed like randomized experiments: An application to postmenopausal hormone therapy and coronary heart disease. Epidemiology. 2008;19: doi: /ede.0b013e e Hernan MA, Hernandez-Diaz S. Beyond the intention-to-treat in comparative effectiveness research. Clin Trials. 2012;9: doi: / Prentice RL, Pettinger M, Anderson GL. Statistical issues arising in the women's health initiative. Biometrics. 2005;61: ; discussion doi: /j x _1.x 19

20 /CIRCULATIONAHA Downloaded from by guest on July 5, Neal B, Perkovic V, Mahaffey KW, Fulcher G, Erondu N, Desai M, Shaw W, Law G, Walton MK, Rosenthal N, Zeeuw Dd, Matthews DR, on behalf of the Cpcg. Optimising the analysis strategy for the canvas program a pre-specified plan for the integrated analyses of the canvas and canvas-r trials. Diabetes Obes Metab doi: /dom Sherman RE, Anderson SA, Dal Pan GJ, Gray GW, Gross T, Hunter NL, LaVange L, Marinac-Dabic D, Marks PW, Robb MA, Shuren J, Temple R, Woodcock J, Yue LQ, Califf RM. Real-world evidence - what is it and what can it tell us? New Engl J Med. 2016;375: doi: /nejmsb U.S. Department of Health and Human Services, Food and Drug Administration. Center for Devices and Radiological Health, Center for Biologics Evaluation and Research. Use of real-world evidence to support regulatory decision-making for medical devices. Draft guidance for industry and food and drug administration staff. Food and Drug Administration, Rockville, USA, September 16, ocuments/ucm pdf. Last Accessed: May 12, European Commission. Directorate-General for Health and Food Safety. STAMP Commission Expert Group. Real world evidence. European Commission, Brussels, Belgium, March 10, Last Accessed: May 12, Califf RM, Robb MA, Bindman AB, Briggs JP, Collins FS, Conway PH, Coster TS, Cunningham FE, De Lew N, DeSalvo KB, Dymek C, Dzau VJ, Fleurence RL, Frank RG, Gaziano JM, Kaufmann P, Lauer M, Marks PW, McGinnis n JM, Richards C, Selby JV, Shulkin DJ, Shuren J, Slavitt AM, Smith SR, Washington BV, White PJ, Woodcock J, Woodson J, Sherman RE. Transforming evidence ence generation to support health and health care decisions. ions. New Engl J Med. 2016;375: doi: /nejmsb Nystrom T, Bodegard J, Nathanson D, Thuresson M, Norhammar A, Eriksson JW. Novel oral glucose-lowering drugs compared to insulin are associated with lower risk of allcause mortality, cardiovascular events and severe hypoglycemia in type 2 diabetes patients. Diabetes Obes Metab doi: /dom Scirica BM, Bhatt DL, Braunwald E, Steg PG, Davidson J, Hirshberg B, Ohman P, Frederich R, Wiviott SD, Hoffman EB, Cavender MA, Udell JA, Desai NR, Mosenzon O, McGuire DK, Ray KK, Leiter LA, Raz I, Committee S-TS, Investigators. Saxagliptin and cardiovascular outcomes in patients with type 2 diabetes mellitus. New Engl J Med. 2013;369: doi: /nejmoa Heerspink HJ, Perkins BA, Fitchett DH, Husain M, Cherney DZ. Sodium glucose cotransporter 2 inhibitors in the treatment of diabetes mellitus: Cardiovascular and kidney effects, potential mechanisms, and clinical applications. Circulation. 2016;134: doi: /circulationaha Marx N, McGuire DK. Sodium-glucose cotransporter-2 inhibition for the reduction of cardiovascular events in high-risk patients with diabetes mellitus. Eur HeartJ. 2016;37: doi: /eurheartj/ehw110 20

21 /CIRCULATIONAHA Sattar N, McLaren J, Kristensen SL, Preiss D, McMurray JJ. Sglt2 inhibition and cardiovascular events: Why did empa-reg outcomes surprise and what were the likely mechanisms? Diabetologia. 2016;59: doi: /s x Downloaded from by guest on July 5,

22 /CIRCULATIONAHA Table 1. Baseline characteristics for all countries combined Downloaded from by guest on July 5, 2017 SGLT-2 inhibitor (N=154,528) Other GLD (N=154,528) n (%) n (%) Age, years, mean (SD) 56.9 (10.0) 57.0 (10.6) Women 68,420 (44.3) 68,772 (44.5) Established cardiovascular disease* 20,044 (13.0) 20,302 (13.1) Acute myocardial infarction 3793 (2.5) 3882 (2.5) Unstable angina 2529 (1.6) 2568 (1.7) Heart failure 4714 (3.1) 4759 (3.1) Atrial fibrillation 5632 (3.6) 5698 (3.7) Stroke 6337 (4.1) 6394 (4.1) Peripheral arterial disease 5239 (3.4) 5229 (3.4) Microvascular disease 42,217 (27.3) 42,215 (27.3) Chronic kidney disease 3920 (2.5) 4171 (2.7) Frailty (yes) 11,982 (7.8) 12,731 (8.2) Baseline glucose-lowering lowering therapies MET 121,500 (78.6) 123,432 (79.9) 9) SU 59,406 (38.4) 59,788 (38.7) DPP-4 inhibitor 51,400 (33.3) 50,088 (32.4) TZD 13,650 (8.8) 12,970 (8.4) GLP-1 receptor agonist 31,355 (20.3) 27,088 (17.5) Insulin 45,573 (29.5) 45,097 (29.2) Cardiovascular therapies es Antihypertensive therapy 123,696 (80.0) 123,563 (80.0) 0) Loop diuretics 14,280 (9.2) 14,314 (9.3) Thiazides 42,446 (27.5) 42,510 (27.5) ACE inhibitors 66,812 (43.2) 67,067 (43.4) ARBs 48,718 (31.5) 48,443 (31.4) Statin therapy 103,968 (67.3) 104,128 (67.4) Index year (0.0) 21,286 (13.8) 71,070 (46.0) 58,951 (38.1) 270 (0.2) 25,713 (16.6) 58,793 (38.0) 66,496 (43.0) *MI, unstable angina, stroke, heart failure, transient ischemic attack, coronary revascularization or within 1 year prior to the index date; Includes angiotensin converting enzyme inhibitors, angiotensin receptor -blockers, thiazides; Data are n (%) unless otherwise stated; DPP- 4=dipeptidyl peptidase-4; egfr=estimated Glomerular Filtration Rate; GLD=glucose-lowering drug; GLP-1=Glucagon-like peptide-1; MET=metformin; SGLT-2=sodium-glucose cotransporter -2; SU=sulfonylurea; TZD=thiazolidinedione. 22

23 /CIRCULATIONAHA Figure Legends Figure 1. Patient flow-chart for all countries/databases combined. A large number of patients were excluded from the other GLD group due to the protocol mandated 1:1 match, and given the smaller number of patients in the SGLT-2i group. GLD=glucose-lowering drug; SGLT-2i=sodium-glucose cotransporter-2 inhibitor. Figure 2. Contribution of the SGLT-2 inhibitor class as a proportion of exposure time in Downloaded from by guest on July 5, 2017 the propensity-match cohorts. A. All countries combined. B. US only. C. European countries combined. HHF=hospitalization for heart failure; SGLT-2=sodium-glucose cotransporter-2. Figure 3. Hazard ratios and 95% CI for the outcome of HHF. A. On treatment, tmen ent, unadjusted. uste d B. On treatment, adjusted (model adjusted for history of heart failure, age, gender, frailty, history of myocardial infarction, history of atrial fibrillation, illal hypertension, obesity/body b mass index, - -blocker use, Ca+-channel blocker use, loop diuretic use, thiazide diuretic use). C. Intent-to-treat, unadjusted CPRD=Clinical Practice Research Datalink; DPV=Diabetes Patientenverlaufsdokumentation; SGLT-2i=sodium-glucose cotransporter-2 inhibitor; THIN=The Health Improvement Network; UK=United Kingdom; US=United States. 23

24 /CIRCULATIONAHA Figure 4. Hazard ratios and 95% CI for the outcome of all-cause death and composite of hospitalization for heart failure or all-cause death. A. All-cause death: on treatment, unadjusted. B. All-cause death: on treatment, adjusted (model adjusted for history of heart failure, age, gender, frailty, history of myocardial infarction, history of atrial fibrillation, - -blocker use, Ca+-channel blocker use, loop diuretic use, thiazide diuretic use). C. All-cause death: intent-to-treat, unadjusted. D. Hospitalization for heart failure or all-cause death: on treatment, unadjusted. E. Hospitalization for heart failure or all-cause death: on Downloaded from by guest on July 5, 2017 treatment, adjusted (model adjusted for history of heart failure, age, gender, frailty, history of myocardial infarction, history of atrial fibrillation, hypertension, obesity/body mass index, duration of diabetes, ACE inhibit - -blocker - -blocker use, Ca+-channel blocker use, loop diuretic use, thiazide diuretic use). F. Hospitalization for heart failure or allcause death: intent-to-treat, t-to-tre unadjusted. CPRD=Clinical Practice Research Datalink; DPV= Diabetes Patientenverlaufsdokumentation; nverlaufsdokumum entation; SGLT-2=sodium-glucose cotransporter-2; THIN=The Th Health Improvement Network; UK=United Kingdom; US=United States. 24

25 Downloaded from by guest on July 5, ,392,254 new users of SGLT-2i or other GLD fulfilling the eligibility criteria 166,033 SGLT-2i 1,226,221 other GLD 11,505 excluded (7%) 1,071,693 excluded (87%) 154,528 SGLT-2i 154,528 other GLD

26 A. Downloaded from by guest on July 5, 2017 B. C. Proportion of exposure time Proportion of exposure time Proportion of exposure time 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% n=154,507 n=107,805 n=107, % 6.7 % 5.6 % 41.8 % 51.0 % 49.1 % 52.7 % 42.3 % 45.3 % HHF All-cause death HHF+All-cause death Canagliflozin Dapagliflozin Empagliflozin n=116,899 n=71,632 n=71, % 5.3 % 5.3 % 19.0 % 19.3 % 19.3 % 75.9 % 75.4 % 75.4 % HHF All-cause death HHF+All-cause death Canagliflozin Dapagliflozin Empagliflozin n=37,608 n=36,173 n=36, % 8.3 % 6.1 % 91.9 % 90.1 % 92.2 % 1.8 % 1.5 % 1.7 % HHF All-cause death HHF+All-cause death Canagliflozin Dapagliflozin Empagliflozin

27 /circ.ahajournals.org/ by guest on July 5, 2017

28 /circ.ahajournals.org/ by guest on July 5, 2017

29 /circ.ahajournals.org/ by guest on July 5, 2017

30 /circ.ahajournals.org/ by guest on July 5, 2017

31 /circ.ahajournals.org/ by guest on July 5, 2017

32 /circ.ahajournals.org/ by guest on July 5, 2017

33 /circ.ahajournals.org/ by guest on July 5, 2017

34 /circ.ahajournals.org/ by guest on July 5, 2017

35 /circ.ahajournals.org/ by guest on July 5, 2017

36 Downloaded from by guest on July 5, 2017 Lower Risk of Heart Failure and Death in Patients Initiated on SGLT-2 Inhibitors Versus Other Glucose-Lowering Drugs: The CVD-REAL Study Mikhail Kosiborod, Matthew A. Cavender, Alex Z. Fu, John P. Wilding, Kamlesh Khunti, Reinhard W. Holl, Anna Norhammar, Kåre I. Birkeland, Marit Jørgensen, Marcus Thuresson, Niki Arya, Johan Bodegård, Niklas Hammar, Peter Fenici and on behalf of the CVD-REAL Investigators and Study Group on behalf of the CVD-REAL Investigators and Study Group Circulation. published online May 18, 2017; Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX Copyright 2017 American Heart Association, Inc. All rights reserved. Print ISSN: Online ISSN: The online version of this article, along with updated information and services, is located on the World Wide Web at: Free via Open Access Data Supplement (unedited) at: Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: Subscriptions: Information about subscribing to Circulation is online at:

37 SUPPLEMENTAL MATERIAL 1

38 Table of Contents CVD-REAL Investigator and Study Group 4 Characteristics of the databases 5 Statistical power calculations 8 SUPPLEMENTARY TABLES Table S1. ICD and Read codes for Type 2 diabetes, Type 1 diabetes and gestational diabetes 9 Table S2. Read codes for history of cardiovascular events 17 Table S3. Comparison of post-propensity match baseline characteristics between the US Truven MarketScan mortality subset and US Truven MarketScan total population 36 Table S4. List of variables used to develop propensity score 38 Table S5. Baseline characteristics for all countries/databases pre-match 40 Table S6. Baseline characteristics for all countries/databases post-match 42 Table S7: Standardized differences in baseline characteristics between SGLT-2 inhibitor and other GLD treatment groups pre- and post-propensity match by country 44 Table S8. Composition of SGLT-2 inhibitor class in propensity matched cohorts 48 Table S9. Index glucose-lowering medication classes for patients in the other GLD group: all countries combined (HHF analysis) 49 Table S10. Index glucose-lowering medication classes for patients in the other GLD group: all countries combined (all-cause death analysis and HHF or all-cause death analysis) 50 Table S11: Number of patients, person-years at risk (on treatment), events and events/100 person-years (incidence rates) for each of the endpoints 51 Table S12. Pooled event rates by treatment groups 52 Table S13. Mean follow-up time (in days) by country and endpoint 53 Table S14. Sensitivity analyses examining association between treatment with SGLT-2i vs. ogld and outcomes of HHF in Sweden, among patients with both in- and outpatient hospital visits with primary diagnosis of heart failure, and those with only inpatient hospital visits with the primary diagnosis of heart failure 54 SUPPLEMENTARY FIGURES Figure S1: Patient selection flow-charts for each country 55 A. US Truven MarketScan 2

39 B. Norway National Registers C. Denmark National Registers D. Sweden National Registers E. UK CPRD/THIN F. Germany Figure S2: Propensity score distribution by country pre-match 58 A. US Truven MarketScan B. Norway National Registers C. Denmark National Registers D. Sweden National Registers E. UK CPRD/THIN F. Germany Figure S3: Propensity score distribution by country post-match 59 A. US Truven MarketScan B. Norway National Registers C. Denmark National Registers D. Sweden National Registers E. UK CPRD/THIN F. Germany Figure S4: Standardized differences between SGLT-2 inhibitor and other GLD treatment groups by country 60 A. US Truven MarketScan B. Norway National Registers C. Denmark National Registers D. Sweden National Registers E. UK CPRD/THIN F. Germany Figure S5: Stepwise sensitivity analysis (sequentially removing comparators): Outcome of hospitalization for heart failure 66 A. TZD removed B. TZD and insulin removed C. TZD, insulin and SU removed Figure S6: Outcome of hospitalization for heart failure excluding patients with baseline GLP-1 RAs 69 Figure S7: Outcomes of hospitalization for heart failure, all-cause death, and hospitalization for heart failure or all-cause for the SGLT-2 inhibitor versus other GLD treatment groups: 70 A. US only B. European countries combined References for the supplementary appendices 71 3

40 CVD-REAL Investigator and Study Group Executive Scientific Committee (Academic Members and Investigators): Mikhail Kosiborod, MD, Saint Luke's Mid America Heart Institute and University of Missouri-Kansas City, Kansas City, MO, United States; Matthew A. Cavender, MD MPH, University of North Carolina, Chapel Hill, NC, United States; Alex Z. Fu, PhD, Georgetown University Medical Center, Washington DC, United States; John P. Wilding MD, PhD, University of Liverpool, Liverpool, United; Kingdom; Kamlesh Khunti, MD PhD, University of Leicester, Leicester, United Kingdom; Anna Norhammar, MD, Cardiology Unit, Department of Medicine, Karolinska Institutet, Stockhom, Sweden; Kåre Birkeland, MD PhD, Department of Endocrinology, University of Oslo, Norway; Marit Eika Jørgensen, MD PhD, cand.med, Steno Diabetes Center, Gentofte, Denmark; Reinhard W. Holl MD PhD, Institute of Epidemiology, University of Ulm, Ulm, Germany; Executive Scientific Committee (AstraZeneca Members): Niklas Hammar, PhD, Senior Director Epidemiology, AZ Gothenburg, Sweden and Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden (Study Principal Investigator); Peter Fenici, MD, PhD, Global Medical Affairs Senior Leader, AZ, Academy House, Cambridge, United Kingdom (Study Leader); Niki Arya, MSc, Principal Statistician, Biometrics and Information Sciences, AZ Gaithersburg, MD, United States (Study Statistician); Kyle Nahrebne, MSc, Global Publication Leader, AZ Gaithersburg, MD, United (Study Publication Leader); All Members of the Executive Steering Committee have contributed to data interpretation and approved this manuscripts analyses and conclusions. Executive Scientific Committee Academic Members MK, MAC AZF, JPW, KK, AN have also reviewed and input the study protocol and amendment, together with the Executive Scientific Committee AZ Members NH, PF, NA, KN and with the Study Core Team Members, KAS, JB, BTB, SED, KB, MFS. Study meta analyses have been conducted by Niki Arya (AstraZeneca) and Marcus Thuresson (Statisticon). Study Core Team (AstraZeneca Members): Karolina Andersson-Sundell, MSc PhD, Assoc.Dir. Epidemiology, MEOR, AZ Gothenburg, Sweden; Johan Bodegård, MD PhD, Medical Evidence Scientific Lead Nordics/Baltic, AZ Oslo, Norway; Betina T. Blak, MSc PhD RWE Scientific Leader, AZ Luton, United Kingdom; Sara E. Dempster, PhD Associate Principal Informatics Scientist, AZ Waltham, MA, United States; Kelly Bell PharmD, MSPhr, HEOR Director, AZ Wilmington, DE, United States; Eric T. Wittbrodt, PharmD, MPH, HEOR Director, AZ Wilmington, DE, United States; Markus F. Scheerer, MSc PhD, Scientific Advisor Diabetes, AZ Wedel, Germany All Study Core Team Members have contributed to country level analyses and local coordination with databases analysts in their respective country. Study Core Team Members, KAS, JB, BTB, SED, KB, MFS have also reviewed and input the study protocol design and amendments. External Investigators and Analysts: Hanne Løvdal Gulseth, MD, PhD, Department of Endocrinology, Oslo University Hospital, Aker, Norway; Bendix Carstensen, PhD, Diabetes Center, Gentofte, Denmark; Marcus Thuresson, PhD Statistician (Study Statistician), Statisticon AB, Uppsala, Sweden; Esther Bollow, Institute of Epidemiology, University of Ulm, Ulm, Germany; Luis Alberto García Rodríguez, MD, CEIFE - Centro Español de Investigación Farmacoepidemiológica, Madrid, Spain; Lucia Cea Soriano, PharmD, PhD, CEIFE Centro Español de Investigación Farmacoepidemiológica and Department of Preventive Medicine and Public Health, Faculty of Medicine, Complutense University of Madrid, Spain; Oscar Fernándex Cantero, CEIFE Centro Español de Investigación Farmacoepidemiológica; Ellen Riehle, MPH, Senior Research Analyst, Truven Health Analytics, an IBM Company, Ann Arbor, MI; Brian Murphy, MS, Senior Data Analyst, Evidera, Waltham, MA. All other External Investigators and Analysts Members have contributed to country level data analyses, quality check and validation and results interpretation in their respective country. 4

41 Characteristics of the databases Truven Health MarketScan Commercial Claims and Encounters (Commercial) and Medicare Supplemental and Coordination of Benefits (Medicare Supplemental) databases comprise enrolment information, demographic information and inpatient medical, outpatient medical and outpatient pharmacy claims data from over 300 large self-insured US employers and over 25 US health plans 1. These study databases satisfy the conditions set forth in Sections (a)-(b)1ii of the Health Insurance Portability and Accountability Act of 1996 privacy rule regarding the determination and documentation of statistically de-identified data Thus, the study did not require external IRB or ethics review. The Truven MarketScan database has been used for comparative effectiveness research since 1989, with over 1100 peer-reviewed publications overall and is considered to be nationally representative in the US for the included patient populations. 2 Truven MarketScan is highly representative of the US commercially insured population aged <65 years, without contributing any specific bias to create differences from the commercially insured population as a whole. Comparisons of age, gender, and region for Truven MarketScan with US Census data (2010) demonstrate close similarities between the two datasets (data on file). Clinical Practice Research Datalink (CPRD) holds anonymized longitudinal primary care patient records collected from over 670 general practices across the UK, covering >11.3 million patients. It includes diagnoses, issued drug prescriptions, clinical measures taken within the general practice, lab tests and referrals to specialist care. 3 Hospitalisation information and specialist care notes are generally recorded by the general practitioner into the primary care patient records. Of the UK practices included in CPRD, 58% are linked to the Hospital Episode Statistics (HES) dataset 3 with detailed hospitalisation information (excluding drug use) on all hospitalisation episodes in England, and to death certificates from the Office of National Statistics (ONS) to derive estimates of all-cause and cause-specific death. CPRD includes approximately 6.9% of the UK population, and patients are broadly representative of the UK general population. 3 Herrett et al 3 assesses the representativeness of CPRD by comparing the age and sex distribution within CPRD to the UK Census in They found that CPRD was broadly representative of the UK population with respect to age and sex. In addition, Herrett et al published a literature review assessing the validation and validity of diagnoses in CPRD (formerly known as GPRD). They concluded that overall, estimates of validity were high. A detailed description of CPRD can be found in a paper by Williams et al. 5 This generalizability of CPRD to the UK population has resulted in CPRD being used in over 1000 publications. 3 The Health Improvement Network (THIN) includes data from >580 UK practices, with similar data to CPRD. 6 THIN coverage of the UK population by 2013 was 5.67% 7 and the representativeness of THIN was described by Blak et al by comparing the distribution of deprivation, morbidities (Quality and Outcomes Framework [QOF] conditions) and demographics to national statistics and national QOF 2006/2007 data. 8 They demonstrated that THIN is representative of the UK population with respect to morbidities, demographics, and mortality rates. Furthermore, in THIN, it was found that between 1990 and 2009 the standardized mortality ratio ranged from 0.81 (95% CI: ; 1990) to 0.93 (95% CI: ; 1995). Adjusting for demographics/deprivation, the 5

42 2006 THIN death rate was 9.08/1000 population, which is consistent with the national death rate of 9.4/1000 population. When Blak et al. was originally published, THIN included 532 general practices. Currently there are approximately 600 general practices. A description of the development of THIN is described in Bourke et al. 9 Lewis et al validated THIN by comparing drug-disease associations found in the original GPRD practices included in THIN, to drug-disease associations found in the new practices included in THIN. 10 They concluded that: THIN data that are collected outside of the GPRD appear as valid as the data collected as part of the GPRD. Other validation studies of THIN are also available, 11, 12 and similar to CPRD, THIN has been used for numerous publications (~400 to date; some of which compare THIN data results to external measures, demonstrating similarities Some of the general practices providing data to CPRD are also providing data to THIN. Therefore, there is an overlap and a potential to combine the two databases in order to increase patient numbers. A recent study within diabetes has shown that 61% of CPRD patients are also included in THIN. 6 This overlap in patients has been taken into account in the present study so that no individual is double counted. The duplicated practices list between CPRD and THIN was obtained by looking for coincidences of patients by practices between both databases. The parameters used for the match were sex, year of birth, family id, and date and Read code of diabetes. 7, 19 For the analyses all THIN practices were retained and the non-overlapping CPRD HES and ONS linked practices. In CPRD, HHF was identified in linked HES with International Statistical Classification of Diseases and Related Health Problems (ICD-system) 10 codes I50.x and all-cause death was identified in linked ONS death data. HHF and all-cause death was identified in the general practice medical records in THIN. For HFF, free text from THIN was collected within days of event and reviewed to ascertain and validate the cause of hospitalisation and the event date. The reviewers were blinded to the index medication. Patients were selected from THIN and/or CPRD from November 2012 to September 2015 (THIN) and to January 2016 (CPRD). This study was approved by the Scientific Review Committee (SRC) of THIN; protocol approval number: 16THIN027A1. Further, the Independent Scientific Advisory Committee (ISAC) of CPRD approved it; protocol approval number: 16_064RAR. The Diabetes-Patienten-Verlaufsdokumentation (DPV; or Diabetes Prospective Follow-up ) initiative involves more than 400 clinical centres predominantly from Germany and Austria, documenting data pertaining to diabetes. Similar to an electronic medical record, relevant data are documented only once and are available for numerous purposes: graphical and tabular description, medical report, treatment plan, diabetes passport, reminder on upcoming visits (watchdog), certification of centers and type 2 diabetes program (DMP) documentation. Every 6 months, anonymized data are sent to the University of Ulm, Germany. 20 Analyses are conducted separately for pediatric and adult patients either anonymised or centre-based after prior written informed consent for regional quality circles. Data of the anonymised DPV registry are used for treatment research in order to investigate practice-oriented questions. A publication list can be found on the website under publications ( Data from this registry have been used extensively in multi-center outcomes 6

43 research, with 452 centers participating. The DPV-initiative is the multicenter benchmark database for patients with diabetes in Germany, includes all levels of diabetes care for the German population of patients with T2D, and has been used in over 370 peer-reviewed publications. 20 The DPV Type 2 diabetes population is consistent with that observed in the national German Health Interview and Examination Surveys 21 when considering the inclusion and exclusion criteria of the CVD-REAL study. In addition, demographic characteristics of the DPV population are consistent with other large data collections of patients with diabetes from Germany, such as the DMP data from Northrhine ( or the DIVE registry ( The National Prescribed Drug Registers in Sweden, Norway and Denmark have full coverage of each country s population. Patients were followed with regard to outcome in the National Patient Register and National Cause of Death Register. The Swedish national database 22 includes information from linkage of three national Swedish registries held by the Swedish National Board of Health and Welfare, with full coverage of the Swedish population: 1) The Prescribed Drug Register July 1, 2005 to December 2016, covering all drug prescriptions filled using Anatomical Therapeutic Chemical codes; 2) The Cause of Death Register ; 3) The National Patient Register covering all hospital admissions and discharge diagnoses in , discharge diagnoses, specialized care and open patient clinic visits in Diagnoses are recorded according to the International Statistical Classification of Diseases and Related Health Problems (ICD) system. Similarly, the Norwegian national database 23 includes type 2 diabetes patient information from three national Norwegian registries with full coverage of the Norwegian population: the Norwegian Prescription Database (July 2004 to July 2016) covering all filled drug prescriptions using ATC codes; the Norwegian Cause of Death Register (1958 to 2014); and the Norwegian Patient Register covering all hospital out-patient clinic visit and discharge diagnoses and all hospital discharge diagnoses for the years 2008 to July Diagnoses are recorded according to the ICD-system. Data linkage is performed by the Norwegian Institute of Public Health. The Danish national database 24 also has a similar structure, and includes T2D patient information from three national Danish registries with full coverage of the Danish population: the Prescribed Drug Register (1990 to 2015) covering all filled drug prescriptions using ATC codes; the Cause of Death Register (1952 to 2014); and the National Patient Register covering all open patient clinic visit diagnoses for 2000 to 2015 and all hospital discharge diagnoses for the years 1980 to 2015, and discharge diagnoses and. Diagnoses are recorded according to the ICD-system. All three registers are held by the Statistics Denmark. Data from Statistics Denmark were made available following an application to Statistics Denmark. The Danish study was approved by the Danish Data Protection Agency (Datatilsynet, registration number ). Data were anonymised, and the requirement for informed consent was therefore waived according to standard analytical procedures with each database owner. 7

44 Statistical power calculations For the primary outcome (HHF) a risk reduction of 20% for SGLT-2i versus ogld was considered clinically meaningful. For 85% power to detect a risk reduction of 20% with a two-sided α-level of 0.05 and a 1:1 treatment allocation (SGLT-2i vs. ogld), a total of 730 events across the matched treatment groups in all the datasets was required. As there were a total of 961 HHF events within the matched cohorts, we had sufficient power to perform the HHF analysis. Analyses were conducted using R- version , 26 in Sweden, Norway and Denmark, STATA version 12.0 (StataCorp LP, College Station, TX, US) in the UK, STATA version 12.0 (StataCorp LP, College Station, TX, US) and SAS version 9.4 (Cary, NC) in the US, and SAS version 9.4 (TS1M1) (Cary, NC, US) in Germany. 8

45 SUPPLEMENTARY TABLES Table S1. ICD and Read codes for Type 2 diabetes, Type 1 diabetes and gestational diabetes Type 2 diabetes ICD codes ICD X0, 250.X2 ICD-10 codes E11 and Read codes 66A4.00 Diabetic on oral treatment 66Ao.00 66At100 66At111 66AV.00 C C C C C C C C C C C C Diabetes type 2 review Type II diabetic dietary review Type 2 diabetic dietary review Diabetic on insulin and oral treatment Diabetes mellitus, adult onset, no mention of complication Maturity onset diabetes Non-insulin dependent diabetes mellitus Diabetes mellitus, adult onset, with ketoacidosis Diabetes mellitus, adult onset, with hyperosmolar coma Diabetes mellitus, adult onset, with ketoacidotic coma Diabetes mellitus, adult onset, with renal manifestation Diabetes mellitus, adult onset, + ophthalmic manifestation Diabetes mellitus, adult onset, + neurological manifestation Diabetes mellitus, adult, + peripheral circulatory disorder Diabetes mellitus, adult with gangrene NIDDM with peripheral circulatory disorder C Non-insulin dependent diabetes mellitus C NIDDM - Non-insulin dependent diabetes mellitus 9

46 C Type 2 diabetes mellitus C Type II diabetes mellitus C C C C C C C C C C C C C C C C C C C C C C C C109A00 C109A11 Non-insulin-dependent diabetes mellitus with renal comps Type II diabetes mellitus with renal complications Type 2 diabetes mellitus with renal complications Non-insulin-dependent diabetes mellitus with ophthalm comps Type II diabetes mellitus with ophthalmic complications Type 2 diabetes mellitus with ophthalmic complications Non-insulin-dependent diabetes mellitus with neuro comps Type II diabetes mellitus with neurological complications Type 2 diabetes mellitus with neurological complications Non-insulin-dependent diabetes mellitus with multiple comps Non-insulin dependent diabetes mellitus with ulcer Type II diabetes mellitus with ulcer Type 2 diabetes mellitus with ulcer Non-insulin dependent diabetes mellitus with gangrene Type II diabetes mellitus with gangrene Type 2 diabetes mellitus with gangrene Non-insulin-dependent diabetes mellitus with retinopathy Type II diabetes mellitus with retinopathy Type 2 diabetes mellitus with retinopathy Non-insulin dependent diabetes mellitus - poor control Type II diabetes mellitus - poor control Type 2 diabetes mellitus - poor control Non-insulin-dependent diabetes mellitus without complication Non-insulin dependent diabetes mellitus with mononeuropathy Type II diabetes mellitus with mononeuropathy 10

47 C109B00 C109B11 C109C00 C109C11 C109C12 C109D00 C109D11 C109D12 C109E00 C109E11 C109E12 C109F00 C109F11 C109F12 C109G00 C109G11 C109G12 C109H00 C109H11 C109H12 C109J00 C109J11 C109J12 C109K00 C10C.11 Non-insulin dependent diabetes mellitus with polyneuropathy Type II diabetes mellitus with polyneuropathy Non-insulin dependent diabetes mellitus with nephropathy Type II diabetes mellitus with nephropathy Type 2 diabetes mellitus with nephropathy Non-insulin dependent diabetes mellitus with hypoglyca coma Type II diabetes mellitus with hypoglycaemic coma Type 2 diabetes mellitus with hypoglycaemic coma Non-insulin depend diabetes mellitus with diabetic cataract Type II diabetes mellitus with diabetic cataract Type 2 diabetes mellitus with diabetic cataract Non-insulin-dependent d m with peripheral angiopath Type II diabetes mellitus with peripheral angiopathy Type 2 diabetes mellitus with peripheral angiopathy Non-insulin dependent diabetes mellitus with arthropathy Type II diabetes mellitus with arthropathy Type 2 diabetes mellitus with arthropathy Non-insulin dependent d m with neuropathic arthropathy Type II diabetes mellitus with neuropathic arthropathy Type 2 diabetes mellitus with neuropathic arthropathy Insulin treated Type 2 diabetes mellitus Insulin treated non-insulin dependent diabetes mellitus Insulin treated Type II diabetes mellitus Hyperosmolar non-ketotic state in type 2 diabetes mellitus Maturity onset diabetes in youth C10D.00 Diabetes mellitus autosomal dominant type 2 C10D.11 Maturity onset diabetes in youth type 2 11

48 C10F.00 C10F.11 C10F000 C10F011 C10F100 C10F111 C10F200 C10F211 C10F300 C10F311 C10F400 C10F411 C10F500 C10F511 C10F600 C10F611 C10F700 C10F711 C10F900 C10F911 C10FA00 C10FA11 C10FB00 C10FB11 C10FC00 C10FC11 C10FD00 Type 2 diabetes mellitus Type II diabetes mellitus Type 2 diabetes mellitus with renal complications Type II diabetes mellitus with renal complications Type 2 diabetes mellitus with ophthalmic complications Type II diabetes mellitus with ophthalmic complications Type 2 diabetes mellitus with neurological complications Type II diabetes mellitus with neurological complications Type 2 diabetes mellitus with multiple complications Type II diabetes mellitus with multiple complications Type 2 diabetes mellitus with ulcer Type II diabetes mellitus with ulcer Type 2 diabetes mellitus with gangrene Type II diabetes mellitus with gangrene Type 2 diabetes mellitus with retinopathy Type II diabetes mellitus with retinopathy Type 2 diabetes mellitus - poor control Type II diabetes mellitus - poor control Type 2 diabetes mellitus without complication Type II diabetes mellitus without complication Type 2 diabetes mellitus with mononeuropathy Type II diabetes mellitus with mononeuropathy Type 2 diabetes mellitus with polyneuropathy Type II diabetes mellitus with polyneuropathy Type 2 diabetes mellitus with nephropathy Type II diabetes mellitus with nephropathy Type 2 diabetes mellitus with hypoglycaemic coma 12

49 C10FD11 C10FE00 C10FE11 C10FF00 C10FF11 C10FG00 C10FG11 C10FH00 C10FJ00 C10FJ11 C10FK00 C10FL00 C10FL11 C10FM00 C10FM11 C10FN00 C10FP00 C10FQ00 C10FR00 C10K.00 C10K000 C10z100 L Type II diabetes mellitus with hypoglycaemic coma Type 2 diabetes mellitus with diabetic cataract Type II diabetes mellitus with diabetic cataract Type 2 diabetes mellitus with peripheral angiopathy Type II diabetes mellitus with peripheral angiopathy Type 2 diabetes mellitus with arthropathy Type II diabetes mellitus with arthropathy Type 2 diabetes mellitus with neuropathic arthropathy Insulin treated Type 2 diabetes mellitus Insulin treated Type II diabetes mellitus Hyperosmolar non-ketotic state in type 2 diabetes mellitus Type 2 diabetes mellitus with persistent proteinuria Type II diabetes mellitus with persistent proteinuria Type 2 diabetes mellitus with persistent microalbuminuria Type II diabetes mellitus with persistent microalbuminuria Type 2 diabetes mellitus with ketoacidosis Type 2 diabetes mellitus with ketoacidotic coma Type 2 diabetes mellitus with exudative maculopathy Type 2 diabetes mellitus with gastroparesis Type A insulin resistance Type A insulin resistance without complication Diabetes mellitus, adult onset, + unspecified complication Pre-existing diabetes mellitus, non-insulin-dependent Type 1 diabetes ICD codes ICD-9 codes 250.x1, 250.X3 ICD-10 codes E10 and O24 13

50 Read codes 66An.00 C108D11 66An.00 66At000 66At011 C C C C C C C C C Diabetes type 1 review Type I diabetes mellitus with nephropathy Diabetes type 1 review Type I diabetic dietary review Type 1 diabetic dietary review Insulin dependent diabetes mellitus Diabetes mellitus, juvenile type, with ketoacidosis Diabetes mellitus, juvenile type, with hyperosmolar coma Diabetes mellitus, juvenile type, with ketoacidotic coma Diabetes mellitus, juvenile type, with renal manifestation Diabetes mellitus, juvenile type, + ophthalmic manifestation Diabetes mellitus, juvenile, + neurological manifestation Diabetes mellitus, juvenile +peripheral circulatory disorder IDDM with peripheral circulatory disorder C Insulin dependent diabetes mellitus C IDDM-Insulin dependent diabetes mellitus C Type 1 diabetes mellitus C Type I diabetes mellitus C C C C C C C C Insulin-dependent diabetes mellitus with renal complications Type I diabetes mellitus with renal complications Type 1 diabetes mellitus with renal complications Insulin-dependent diabetes mellitus with ophthalmic comps Type 1 diabetes mellitus with ophthalmic complications Insulin-dependent diabetes mellitus with neurological comps Type I diabetes mellitus with neurological complications Type 1 diabetes mellitus with neurological complications 14

51 C C C C C C C C C C C C C C C C C C108A00 C108A11 C108B00 C108B11 C108C00 C108D00 C108D11 C108E00 C108E11 C108E12 Insulin dependent diabetes mellitus with multiple complicatn Unstable insulin dependent diabetes mellitus Unstable type I diabetes mellitus Unstable type 1 diabetes mellitus Insulin dependent diabetes mellitus with ulcer Type I diabetes mellitus with ulcer Type 1 diabetes mellitus with ulcer Insulin dependent diabetes mellitus with gangrene Insulin dependent diabetes mellitus with retinopathy Type I diabetes mellitus with retinopathy Type 1 diabetes mellitus with retinopathy Insulin dependent diabetes mellitus - poor control Type I diabetes mellitus - poor control Type 1 diabetes mellitus - poor control Insulin dependent diabetes maturity onset Type I diabetes mellitus maturity onset Type 1 diabetes mellitus maturity onset Insulin-dependent diabetes without complication Type I diabetes mellitus without complication Insulin dependent diabetes mellitus with mononeuropathy Type I diabetes mellitus with mononeuropathy Insulin dependent diabetes mellitus with polyneuropathy Insulin dependent diabetes mellitus with nephropathy Type I diabetes mellitus with nephropathy Insulin dependent diabetes mellitus with hypoglycaemic coma Type I diabetes mellitus with hypoglycaemic coma Type 1 diabetes mellitus with hypoglycaemic coma 15

52 C108F00 C108F11 C108G00 C108H00 C108H11 C108J00 C108J11 C108J12 Insulin dependent diabetes mellitus with diabetic cataract Type I diabetes mellitus with diabetic cataract Insulin dependent diab mell with peripheral angiopathy Insulin dependent diabetes mellitus with arthropathy Type I diabetes mellitus with arthropathy Insulin dependent diab mell with neuropathic arthropathy Type I diabetes mellitus with neuropathic arthropathy Type 1 diabetes mellitus with neuropathic arthropathy C10C.12 Maturity onset diabetes in youth type 1 C10E.00 C10E.11 C10E.12 C10E000 C10E012 C10E100 C10E111 C10E112 C10E200 C10E212 C10E300 C10E311 C10E312 C10E400 C10E411 C10E412 C10E500 C10E511 Type 1 diabetes mellitus Type I diabetes mellitus Insulin dependent diabetes mellitus Type 1 diabetes mellitus with renal complications Insulin-dependent diabetes mellitus with renal complications Type 1 diabetes mellitus with ophthalmic complications Type I diabetes mellitus with ophthalmic complications Insulin-dependent diabetes mellitus with ophthalmic comps Type 1 diabetes mellitus with neurological complications Insulin-dependent diabetes mellitus with neurological comps Type 1 diabetes mellitus with multiple complications Type I diabetes mellitus with multiple complications Insulin dependent diabetes mellitus with multiple complicat Unstable type 1 diabetes mellitus Unstable type I diabetes mellitus Unstable insulin dependent diabetes mellitus Type 1 diabetes mellitus with ulcer Type I diabetes mellitus with ulcer 16

53 C10E512 C10E600 C10E611 C10E700 C10E711 C10E712 C10E800 C10E812 C10E900 C10E911 C10E912 C10EA00 C10EA11 C10EA12 C10EB00 C10EC00 C10EC11 C10EC12 C10ED00 C10ED12 C10EE00 C10EE12 C10EF00 C10EF12 C10EG00 C10EH00 C10EJ00 Insulin dependent diabetes mellitus with ulcer Type 1 diabetes mellitus with gangrene Type I diabetes mellitus with gangrene Type 1 diabetes mellitus with retinopathy Type I diabetes mellitus with retinopathy Insulin dependent diabetes mellitus with retinopathy Type 1 diabetes mellitus - poor control Insulin dependent diabetes mellitus - poor control Type 1 diabetes mellitus maturity onset Type I diabetes mellitus maturity onset Insulin dependent diabetes maturity onset Type 1 diabetes mellitus without complication Type I diabetes mellitus without complication Insulin-dependent diabetes without complication Type 1 diabetes mellitus with mononeuropathy Type 1 diabetes mellitus with polyneuropathy Type I diabetes mellitus with polyneuropathy Insulin dependent diabetes mellitus with polyneuropathy Type 1 diabetes mellitus with nephropathy Insulin dependent diabetes mellitus with nephropathy Type 1 diabetes mellitus with hypoglycaemic coma Insulin dependent diabetes mellitus with hypoglycaemic coma Type 1 diabetes mellitus with diabetic cataract Insulin dependent diabetes mellitus with diabetic cataract Type 1 diabetes mellitus with peripheral angiopathy Type 1 diabetes mellitus with arthropathy Type 1 diabetes mellitus with neuropathic arthropathy 17

54 C10EK00 C10EL00 C10EL11 C10EM00 C10EM11 C10EN00 C10EN11 C10EP00 C10EP11 C10EQ00 C10z000 L Hj3.00 8Hj4.00 8Hj5.00 Type 1 diabetes mellitus with persistent proteinuria Type 1 diabetes mellitus with persistent microalbuminuria Type I diabetes mellitus with persistent microalbuminuria Type 1 diabetes mellitus with ketoacidosis Type I diabetes mellitus with ketoacidosis Type 1 diabetes mellitus with ketoacidotic coma Type I diabetes mellitus with ketoacidotic coma Type 1 diabetes mellitus with exudative maculopathy Type I diabetes mellitus with exudative maculopathy Type 1 diabetes mellitus with gastroparesis Diabetes mellitus, juvenile type, + unspecified complication Pre-existing diabetes mellitus, insulin-dependent Referral to DAFNE diabetes structured education programme Referral to DESMOND diabetes structured education programme Referral to XPERT diabetes structured education programme 8I82.00 Did not complete DAFNE diabetes structured education program 8I83.00 Did not complete DESMOND diabetes structured educat program 8I84.00 Did not complete XPERT diabetes structured education program 9NiC.00 9NiD.00 9NiE.00 9OLG.00 9OLH.00 9OLJ.00 9OLK.00 9OLL.00 Did not attend DAFNE diabetes structured education programme Did not attend DESMOND diabetes structured education program Did not attend XPERT diabetes structured education programme Attended XPERT diabetes structured education programme Attended DAFNE diabetes structured education programme DAFNE diabetes structured education programme completed DESMOND diabetes structured education programme completed XPERT diabetes structured education programme completed Gestational Diabetes 18

55 ICD codes ICD-9 codes ICD-10:O24.4 Read codes ZC2CB00 ZV13F00 L L CE Ay.00 8CE0000 Q44B.00 66AX.00 Dietary advice for gestational diabetes [V]Personal history of gestational diabetes mellitus Gestational diabetes mellitus Gestational diabetes mellitus Gestational diabetes information leaflet given Gestational diabetes mellitus annual review Gestational diabetes information leaflet given Syndrome of infant of mother with gestational diabetes Diabetes: shared care in pregnancy - diabetol and obstet Diabetic pre-pregnancy counselling L Diabetes mellitus during pregnancy/childbirth/puerperium L L L L L180z00 Diabetes mellitus - unspec whether in pregnancy/puerperium Diabetes mellitus during pregnancy - baby delivered Diabetes mellitus during pregnancy - baby not yet delivered Diabetes mellitus arising in pregnancy Diabetes mellitus in pregnancy/childbirth/puerperium NOS 19

56 Table S2. Read codes for history of cardiovascular events Myocardial infarction G G Acute myocardial infarction Cardiac rupture following myocardial infarction (MI) G Acute anterolateral infarction G Other specified anterior myocardial infarction G G G G G G301z00 Acute anteroapical infarction Attack - heart Heart attack MI - acute myocardial infarction Acute anteroseptal infarction Anterior myocardial infarction NOS G Acute inferolateral infarction G Acute inferoposterior infarction G Posterior myocardial infarction NOS G Lateral myocardial infarction NOS G True posterior myocardial infarction G Acute subendocardial infarction G Acute non-q wave infarction G Inferior myocardial infarction NOS G Acute Q-wave infarct G30B.00 G30X.00 G30X000 G30y.00 G30y000 G30y100 G30y200 G30yz00 G30z.00 G G Acute posterolateral myocardial infarction Acute transmural myocardial infarction of unspecif site Acute ST segment elevation myocardial infarction Other acute myocardial infarction Acute atrial infarction Acute papillary muscle infarction Acute septal infarction Other acute myocardial infarction NOS Acute myocardial infarction NOS Other acute and subacute ischaemic heart disease Acute coronary syndrome G Postmyocardial infarction syndrome 20

57 G31y100 G31y200 G Microinfarction of heart Subendocardial ischaemia Subsequent myocardial infarction G Subsequent myocardial infarction of anterior wall G Subsequent myocardial infarction of inferior wall G Subsequent myocardial infarction of other sites G35X.00 Subsequent myocardial infarction of unspecified site ECG: old myocardial infarction ECG: subendocardial infarct 323Z.00 ECG: myocardial infarct NOS 889A.00 Diab mellit insulin-glucose infus acute myocardial infarct G Acute non-st segment elevation myocardial infarction G Coronary thrombosis not resulting in myocardial infarction G Certain current complication follow acute myocardial infarct G Haemopericardium/current comp folow acut myocard infarct G Ruptur chordae tendinae/curr comp fol acute myocard infarct G Rupture papillary muscle/curr comp fol acute myocard infarct G Postoperative subendocardial myocardial infarction Gyu3100 Gyu3400 [X]Other current complicatns following acute myocard infarct [X]Acute transmural myocardial infarction of unspecif site Unstable Angina G Crescendo angina G Unstable angina G Angina at rest G G G G Unstable angina Angina at rest Refractory angina Worsening angina G Angina decubitus G G330z00 Nocturnal angina Angina decubitus NOS G Prinzmetal's angina G Variant angina pectoris 21

58 Ischemic Stroke 14A7.00 H/O: CVA/stroke 14A7.11 H/O: CVA 14A7.12 H/O: stroke 14AK M M200 8HBJ.00 G G63y000 G63y100 G G G H/O: Stroke in last year Stroke 6 month review Stroke initial post discharge review Stroke / transient ischaemic attack referral Infarction - precerebral Cerebral infarct due to thrombosis of precerebral arteries Cerebral infarction due to embolism of precerebral arteries CVA - cerebral artery occlusion Infarction - cerebral Stroke due to cerebral arterial occlusion G Cerebral thrombosis G Cerebral infarction due to thrombosis of cerebral arteries G Cerebral embolism G Cerebral embolus G G64z.00 G64z.11 G64z.12 G64z000 G64z100 G64z111 G64z200 G64z300 G64z400 Cerebral infarction due to embolism of cerebral arteries Cerebral infarction NOS Brainstem infarction NOS Cerebellar infarction Brainstem infarction Wallenberg syndrome Lateral medullary syndrome Left sided cerebral infarction Right sided cerebral infarction Infarction of basal ganglia G Multiple and bilateral precerebral artery syndromes G G G G Stroke and cerebrovascular accident unspecified CVA unspecified Stroke unspecified CVA - Cerebrovascular accident unspecified 22

59 G Middle cerebral artery syndrome G Anterior cerebral artery syndrome G Posterior cerebral artery syndrome G Brain stem stroke syndrome G Cerebellar stroke syndrome G Pure motor lacunar syndrome G Pure sensory lacunar syndrome G Left sided CVA G Right sided CVA G Moyamoya disease G Nonpyogenic venous sinus thrombosis G Cereb infarct due cerebral venous thrombosis, nonpyogenic G Cereb autosom dominant arteriop subcort infarcts leukoenceph G Sequelae of cerebral infarction G68X.00 G6W..00 G6X..00 Gyu6300 Gyu6400 Gyu6C00 Gyu6G00 Sequelae of stroke,not specfd as h'morrhage or infarction Cereb infarct due unsp occlus/stenos precerebr arteries Cerebrl infarctn due/unspcf occlusn or sten/cerebrl artrs [X]Cerebrl infarctn due/unspcf occlusn or sten/cerebrl artrs [X]Other cerebral infarction [X]Sequelae of stroke,not specfd as h'morrhage or infarction [X]Cereb infarct due unsp occlus/stenos precerebr arteries L Stroke in the puerperium ZLEP.00 ZV12511 ZV12512 Discharge from stroke serv [V]Personal history of stroke [V]Personal history of cerebrovascular accident (CVA) Hemostroke Evacuation of haematoma from temporal lobe of brain Evacuation of haematoma from cerebellum Evacuation of intracerebral haematoma NEC Aspiration of haematoma of brain tissue G G G Intracerebral haemorrhage CVA - cerebrovascular accid due to intracerebral haemorrhage Stroke due to intracerebral haemorrhage 23

60 G Cortical haemorrhage G Internal capsule haemorrhage G Basal nucleus haemorrhage G Cerebellar haemorrhage G Pontine haemorrhage G Bulbar haemorrhage G External capsule haemorrhage G Intracerebral haemorrhage, intraventricular G Intracerebral haemorrhage, multiple localized G Lobar cerebral haemorrhage G61X.00 G61X000 G61X100 G61z.00 Intracerebral haemorrhage in hemisphere, unspecified Left sided intracerebral haemorrhage, unspecified Right sided intracerebral haemorrhage, unspecified Intracerebral haemorrhage NOS G Sequelae of intracerebral haemorrhage Gyu6200 Gyu6F00 [X]Other intracerebral haemorrhage [X]Intracerebral haemorrhage in hemisphere, unspecified Evacuation of subdural haematoma Drainage of subdural space 7034y z00 Other specified drainage of subdural space Drainage of subdural space NOS G Subdural haemorrhage - nontraumatic G Subdural haematoma - nontraumatic G Subdural haemorrhage NOS S Subdural haemorrhage following injury S Closed traumatic subdural haemorrhage S S S S S S S Subdural h'ge inj no open intracranial wnd + unspec consc Subdural h'ge inj no open intracranial wound+no loss consc Subdural h'ge inj no open intracranial wound+<1hr loss consc Subdural h'ge inj no open intracran wnd+1-24hr loss consc Subdural h'ge inj no open intracranial wnd+>24 LOC +recovery Subdural h'ge inj no open intracran wnd+>24hr LOC -restored Subdural h'ge inj no open intracran wnd+loc unspec duration 24

61 S622z00 Subdural h'ge inj no open intracran wound+concussion unspec S Open traumatic subdural haemorrhage S S S S S S S S623z00 Subdural h'ge inj + open intracranial wound + unspec consc Subdural h'ge inj + open intracranial wound+no loss consc Subdural h'ge inj + open intracranial wound+<1hr loss consc Subdural h'ge inj + open intracranial wnd+1-24hr loss consc Subdural h'ge inj + open intracran wound+>24hr LOC +recovery Subdural h'ge inj + open intracran wnd+>24hr LOC -restored Subdural h'ge inj + open intracran wnd+loc unspec duration Subdural h'ge inj + open intracranial wnd+concussion unspec S Traumatic subdural haemorrhage S Traumatic subdural haematoma S S G Traumatic subdural haematoma without open intracranial wound Traumatic subdural haematoma with open intracranial wound Subarachnoid haemorrhage G Subarachnoid haemorrhage from carotid siphon and bifurcation G Subarachnoid haemorrhage from middle cerebral artery G Subarachnoid haemorrhage from anterior communicating artery G Subarachnoid haemorrhage from posterior communicating artery G Subarachnoid haemorrhage from basilar artery G Subarachnoid haemorrhage from vertebral artery G60X.00 G60z.00 Subarachnoid haemorrh from intracranial artery, unspecif Subarachnoid haemorrhage NOS G Sequelae of subarachnoid haemorrhage Gyu6000 Gyu6100 Gyu6E00 S [X]Subarachnoid haemorrhage from other intracranial arteries [X]Other subarachnoid haemorrhage [X]Subarachnoid haemorrh from intracranial artery, unspecif Subarachnoid haemorrhage following injury S Closed traumatic subarachnoid haemorrhage S S S S Subarachnoid h'ge inj no open intracran wound + unspec consc Subarachnoid h'ge inj no open intracran wnd+no loss consc Subarachnoid h'ge inj no open intracran wnd+<1hr loss consc Subarachnoid h'ge inj no open intracran wound hr LOC 25

62 S S S S620z00 Subarachnoid h'ge inj no open intracran wnd+>24 LOC+recovery Subarach h'ge inj no open intracran wnd+>24hrs LOC-restored Subarach h'ge inj no open intracran wnd+loc unspec duration Subarach h'ge inj no open intracran wnd + concussion unspec S Open traumatic subarachnoid haemorrhage S S S S S S S S621z00 Subarachnoid h'ge inj + open intracran wound + unspec consc Subarachnoid h'ge inj + open intracranial wound + no LOC Subarachnoid h'ge inj + open intracran wound+<1hr loss consc Subarachnoid h'ge inj + open intracran wnd+1-24hr loss consc Subarach h'ge inj + open intracran wnd +>24hr LOC + recovery Subarach h'ge inj + open intracran wnd+>24hr LOC -restored Subarach h'ge inj + open intracran wnd+loc unspec duration Subarachnoid h'ge inj + open intracran wnd+concussion unspec S Traumatic subarachnoid haemorrhage Evacuation of extradural haematoma G Ruptured berry aneurysm G Other and unspecified intracranial haemorrhage G Extradural haemorrhage - nontraumatic G62z.00 Intracranial haemorrhage NOS G Sequelae of other nontraumatic intracranial haemorrhage Gyu6B00 S S S [X]Sequelae of other nontraumatic intracranial haemorrhage Cerebral haemorrhage following injury Extradural haemorrhage following injury Traumatic cerebral haemorrhage S Middle meningeal haemmorhage following injury S Closed traumatic extradural haemorrhage S Epidural haematoma following injury S S S S S S Extradural h'ge inj no open intracranial wnd + unspec consc Extradural h'ge inj no open intracranial wnd + no loss consc Extradural h'ge inj no open intracranial wnd+<1hr loss consc Extradural h'ge inj no open intracran wnd+1-24hr loss consc Extradural h'ge inj no open intracran wnd+>24hr LOC+recovery Extradural h'ge inj no open intracran wnd+>24hr LOC-restored 26

63 S S624z00 Extradural h'ge inj no open intracra wnd+loc unspec duration Extradural h'ge inj no open intracran wnd+concussion unspec S Open traumatic extradural haemorrhage S S S S S S S S625z00 Extradural h'ge inj + open intracranial wnd + unspec consc Extradural h'ge inj + open intracranial wound+no loss consc Extradural h'ge inj + open intracranial wnd+<1hr loss consc Extradural h'ge inj + open intracran wnd+1-24hr loss consc Extradural h'ge inj + open intracran wnd+>24hr LOC+recovery Extradural h'ge inj + open intracran wnd+>24hr LOC -restored Extradural h'ge inj + open intracran wnd+loc unspec duration Extradural h'ge inj + open intracran wnd+concussion unspec S Epidural haemorrhage S62A.00 S62A000 S62A100 S62z.00 S Traumatic extradural haematoma Traumatic extradural haemat without open intracranial wound Traumatic extradural haematoma with open intracranial wound Cerebral haemorrhage following injury NOS Other cerebral haemorrhage following injury S Other cerebral h'ge after injury no open intracranial wound S Cerebral compression due to injury S Intracranial haematoma following injury S S S S S S S S630z00 Oth cerebral h'ge inj no open intracran wnd+unspec consc Oth cerebral h'ge inj no open intracranial wnd+no loss consc Oth cerebral h'ge inj no open intracran wnd+<1hr loss consc Oth cerebral h'ge inj no open intracran wnd+1-24hr LOC Oth cereb h'ge inj no open intracran wnd+>24hr LOC +recovery Oth cereb h'ge inj no open intracran wnd+>24hr LOC -restored Oth cereb h'ge inj no open intracran wnd+loc unspec duration Oth cereb h'ge inj no open intracran wnd+concussion unspec S Other cerebral h'ge after injury + open intracranial wound S S S S Oth cerebral h'ge inj + open intracran wnd + unspec consc Oth cerebral h'ge inj + open intracranial wnd+no loss consc Oth cerebral h'ge inj + open intracran wnd+<1hr loss consc Oth cerebral h'ge inj + open intracran wnd+1-24hr loss consc 27

64 S S S S631z00 S63z.00 Oth cereb h'ge inj + open intracran wnd+>24hr LOC + recovery Oth cereb h'ge inj + open intracran wnd+>24hr LOC -restored Oth cereb h'ge inj + open intracran wnd+loc unspec duration Oth cereb h'ge inj + open intracran wnd+concussion unspec Other cerebral haemorrhage following injury NOS Heart Failure 14A6.00 H/O: heart failure 14AM.00 1O1..00 H/O: Heart failure in last year Heart failure confirmed 388D.00 New York Heart Assoc classification heart failure symptoms 662f.00 New York Heart Association classification - class I 662g.00 New York Heart Association classification - class II 662h.00 New York Heart Association classification - class III 662i.00 New York Heart Association classification - class IV 662p.00 Heart failure 6 month review 662T.00 Congestive heart failure monitoring 662W.00 Heart failure annual review 679X.00 Heart failure education 67D4.00 Heart failure information given to patient 8B29.00 Cardiac failure therapy 8CL3.00 8H2S.00 8HBE.00 8Hg8.00 8HHb.00 8HHz.00 8Hk0.00 8HTL.00 9N0k.00 9N2p.00 9N4s.00 9N6T.00 9Or..00 Heart failure care plan discussed with patient Admit heart failure emergency Heart failure follow-up Discharge from practice nurse heart failure clinic Referral to heart failure nurse Referral to heart failure exercise programme Referred to heart failure education group Referral to heart failure clinic Seen in heart failure clinic Seen by community heart failure nurse Did not attend practice nurse heart failure clinic Referred by heart failure nurse specialist Heart failure monitoring administration 28

65 9Or0.00 9Or1.00 9Or2.00 9Or3.00 9Or4.00 9Or5.00 Heart failure review completed Heart failure monitoring telephone invite Heart failure monitoring verbal invite Heart failure monitoring first letter Heart failure monitoring second letter Heart failure monitoring third letter G Hypertensive heart&renal dis wth (congestive) heart failure G Hyperten heart&renal dis+both(congestv)heart and renal fail G1yz100 G5y4z00 G5yy900 G5yyA00 L09y200 Q48y100 Rheumatic left ventricular failure Post cardiac operation heart failure NOS Left ventricular systolic dysfunction Left ventricular diastolic dysfunction Cardiac failure following abortive pregnancy Congenital cardiac failure Q Neonatal cardiac failure SP08400 SP08500 SP11100 SP11111 G G Heart transplant failure and rejection Heart-lung transplant failure and rejection Cardiac insufficiency as a complication of care Heart failure as a complication of care Heart failure Cardiac failure G Congestive heart failure G Congestive cardiac failure G Right heart failure G Right ventricular failure G Biventricular failure G G G G G Acute congestive heart failure Chronic congestive heart failure Decompensated cardiac failure Compensated cardiac failure Congestive heart failure due to valvular disease G Left ventricular failure G Asthma - cardiac 29

66 G Pulmonary oedema - acute G Impaired left ventricular function G Acute left ventricular failure G Acute heart failure G Heart failure with normal ejection fraction G HFNEF - heart failure with normal ejection fraction G Heart failure with preserved ejection fraction G Right ventricular failure G58z.00 G58z.11 G58z.12 Heart failure NOS Weak heart Cardiac failure NOS Transient ischemic attack 8HBJ.00 Fyu5500 G G G G Stroke / transient ischaemic attack referral [X]Other transnt cerebral ischaemic attacks+related syndroms Transient cerebral ischaemia Drop attack Transient ischaemic attack Vertebro-basilar insufficiency G Basilar artery syndrome G Insufficiency - basilar artery G Vertebral artery syndrome G Vertebro-basilar artery syndrome G Subclavian steal syndrome G Carotid artery syndrome hemispheric G Multiple and bilateral precerebral artery syndromes G Transient global amnesia G Vertebrobasilar insufficiency G65y.00 G65z.00 G65z000 G65z100 G65zz00 Other transient cerebral ischaemia Transient cerebral ischaemia NOS Impending cerebral ischaemia Intermittent cerebral ischaemia Transient cerebral ischaemia NOS Coronary artery bypass graft 30

67 7922z00 Allograft replacement of coronary artery NOS Allograft replacement of four or more coronary arteries Allograft replacement of one coronary artery Allograft replacement of three coronary arteries Allograft replacement of two coronary arteries Autograft replacement of four of more coronary arteries NEC Autograft replacement of one coronary artery NEC Autograft replacement of three coronary arteries NEC Autograft replacement of two coronary arteries NEC 7925z y z y00 Connection of mammary artery to coronary artery NOS Connection of mammary artery to coronary artery OS Connection of other thoracic artery to coronary artery NOS Connection of other thoracic artery to coronary artery OS Coronary artery bypass graft operations Double anastom thoracic arteries to coronary arteries NEC Double anastomosis of mammary arteries to coronary arteries Double implant of mammary arteries into coronary arteries Double implant thoracic arteries into coronary arteries NEC SP z y00 Mechanical complication of coronary bypass Other autograft replacement of coronary artery NOS Other autograft replacement of coronary artery OS 792D.00 Other bypass of coronary artery 792Dz00 Other bypass of coronary artery NOS 792C.00 Other replacement of coronary artery 7922y00 792Dy y00 792Cy y00 Other specified allograft replacement of coronary artery Other specified other bypass of coronary artery Other specified prosthetic replacement of coronary artery Other specified replacement of coronary artery Other specified revision of bypass for coronary artery Percut translum balloon angioplasty bypass graft coronary a 7923z00 Prosthetic replacement of coronary artery NOS Prosthetic replacement of four or more coronary arteries Prosthetic replacement of one coronary artery 31

68 Prosthetic replacement of three coronary arteries Prosthetic replacement of two coronary arteries 792C Cz z00 Replacement of coronary arteries using multiple methods Replacement of coronary artery NOS Revision of bypass for coronary artery NOS Revision of bypass for four or more coronary arteries Revision of bypass for one coronary artery Revision of bypass for three coronary arteries Revision of bypass for two coronary arteries Revision of connection of thoracic artery to coronary artery 7920z y00 Saphenous vein graft replacement coronary artery NOS Saphenous vein graft replacement of coronary artery OS Saphenous vein graft replacement of four+ coronary arteries Saphenous vein graft replacement of one coronary artery Saphenous vein graft replacement of three coronary arteries Saphenous vein graft replacement of two coronary arteries Single anast mammary art to left ant descend coronary art Single anastomosis of mammary artery to coronary artery NEC Single anastomosis of thoracic artery to coronary artery NEC Single implantation of mammary artery into coronary artery Single implantation thoracic artery into coronary artery NEC Transection of muscle bridge of coronary artery Transposition of coronary artery NEC ZV45700 ZV45K00 ZV45K00 ZV45K11 ZV45K11 [V]Presence of aortocoronary bypass graft [V]Presence of coronary artery bypass graft [V]Presence of coronary artery bypass graft [V]Presence of coronary artery bypass graft - CABG [V]Presence of coronary artery bypass graft - CABG LIMA sequential anastomosis RIMA sequential anastomosis LIMA single anastomosis RIMA single anastomosis Perc trans three dimen electroanat mapping conduct sys heart 32

69 Repair of arteriovenous fistula of coronary artery Repair of aneurysm of coronary artery 7927y z00 792B000 Other specified other open operation on coronary artery Other open operation on coronary artery NOS Endarterectomy of coronary artery NEC 792B.00 Repair of coronary artery NEC 792By00 792Bz00 Other specified repair of coronary artery Repair of coronary artery NOS 792y.00 Other specified operations on coronary artery 792z.00 Coronary artery operations NOS Other open operations on coronary artery Other open operations on coronary artery Revision of bypass for coronary artery Prosthetic replacement of coronary artery Prosthetic bypass of coronary artery Other autograft replacement of coronary artery Creation of bypass from mammary artery to coronary artery Connection of mammary artery to coronary artery Connection of other thoracic artery to coronary artery Allograft replacement of coronary artery Percutaneous coronary intervention Open angioplasty of coronary artery Percut transluminal balloon angioplasty one coronary artery Transluminal balloon angioplasty of coronary artery Percut translum balloon angioplasty mult coronary arteries Percutaneous balloon coronary angioplasty 7928y z00 Transluminal balloon angioplasty of coronary artery OS Transluminal balloon angioplasty of coronary artery NOS Percutaneous transluminal laser coronary angioplasty Other therapeutic transluminal operations on coronary artery Percut transluminal coronary thrombolysis with streptokinase Percut translum coronary thrombolytic therapy- streptokinase Percut translum inject therap subst to coronary artery NEC 33

70 Rotary blade coronary angioplasty Insertion of coronary artery stent 7929y z00 ZV45800 ZV45L00 Other therapeutic transluminal op on coronary artery OS Other therapeutic transluminal op on coronary artery NOS [V]Presence of coronary angioplasty implant and graft [V]Status following coronary angioplasty NOS 793G.00 Perc translumin balloon angioplasty stenting coronary artery 793K.00 Transluminal operations internal mammary artery side branch Transluminal heart assist operations Other therapeutic transluminal operations on heart Diagnostic transluminal operations on heart 793Gy00 793Gz00 793Ky00 793Kz00 793G H K G G G y z y z00 OS perc translumina balloon angioplast stenting coronary art Perc translum balloon angioplasty stenting coronary art NOS OS transluminal operations internal mammary art side branch Transluminal operations internal mammary art side branch NOS Perc translum ball angio insert 1-2 drug elut stents cor art Percutaneous transluminal balloon dilation cardiac conduit Translum occlusion left internal mammary artery side branch Perc tran ball angio ins 3 or more drug elut stents cor art Perc translum balloon angioplasty insert 1-2 stents cor art Percutaneous cor balloon angiop 3 more stents cor art NEC Other therapeutic transluminal operation on heart OS Other therapeutic transluminal operation on heart NOS Other specified diagnostic transluminal operation on heart Diagnostic transluminal operation on heart NOS Percut translum cutting balloon angioplasty coronary artery Insertion of drug-eluting coronary artery stent Percutaneous transluminal atherectomy of coronary artery Transluminal insertion of pulsation balloon into aorta Percut translum balloon angioplasty bypass graft coronary a Peripheral artery disease Aorto-iliac disease G Extremity artery atheroma 34

71 G702z00 G G G G G G Extremity artery atheroma NOS Peripheral ischaemic vascular disease Ischaemia of legs Peripheral ischaemia Raynaud's phenomenon Buerger's disease Presenile gangrene G Peripheral gangrene G G Gangrene of toe Gangrene of foot G Ischaemic foot G Peripheral arterial disease G73y.00 G73y000 G73y100 G73yz00 G73z.00 G73z000 G73z011 G73z012 G73z100 G73zz00 G G G G Other specified peripheral vascular disease Diabetic peripheral angiopathy Peripheral angiopathic disease EC NOS Other specified peripheral vascular disease NOS Peripheral vascular disease NOS Intermittent claudication Claudication Vascular claudication Spasm of peripheral artery Peripheral vascular disease NOS Arterial embolism and thrombosis Arterial embolus and thrombosis Thrombosis - arterial Arterial embolic and thrombotic occlusion G Aortoiliac obstruction G Leriche's syndrome G G G G G Embolism and thrombosis of the femoral artery Embolism and thrombosis of the popliteal artery Embolism and thrombosis of the anterior tibial artery Embolism and thrombosis of the dorsalis pedis artery Embolism and thrombosis of the posterior tibial artery 35

72 G G742B00 G742z00 G74y000 G74y100 G74y200 G74y300 G74z.00 Embolism and thrombosis of a leg artery NOS Post radiological embolism of lower limb artery Peripheral arterial embolism and thrombosis NOS Embolism and/or thrombosis of the common iliac artery Embolism and/or thrombosis of the internal iliac artery Embolism and/or thrombosis of the external iliac artery Embolism and thrombosis of the iliac artery unspecified Arterial embolism and thrombosis NOS G Stricture of artery G Necrosis of artery G76A.00 G76z.00 G76z000 G76z100 G76z200 G7y..00 G7z..00 Rango 14NB.00 Arterial insufficiency Disorders of arteries and arterioles NOS Iliac artery occlusion Femoral artery occlusion Popliteal artery occlusion Other specified arterial, arteriole or capillary disease Arterial, arteriole and capillary diseases NOS H/O: Peripheral vascular disease procedure 2I16.00 O/E - gangrene R [D]Gangrene R R HlP.00 7A A A4..00 [D]Gangrene of toe in diabetic [D]Widespdiabetic foot gangrene Referred for peripheral artery disease assessment Bypass bifurc aorta by anastom aorta to femoral artery NEC Bypass bifurcation aorta by anastom aorta to iliac artery Iliac and femoral artery operations 7A41.00 Other bypass of iliac artery 7A41.11 Other bypass of iliac artery by anastomosis 7A A A A41500 Bypass iliac artery by iliac/femoral artery anastomosis NEC Emerg bypass iliac artery by femoral/femoral art anast NEC Bypass iliac artery by femoral/femoral art anastomosis NEC Emerg bypass iliac artery by aorta/ext iliac art anast NEC 36

73 7A A A A A41A00 7A41B00 7A41C00 7A41D00 7A41E00 7A41y00 7A41z00 Emerg bypass leg artery by aorta/com fem art anastomosis NEC Emerg bypass leg artery by aorta/deep fem anastomosis NEC Emerg bypass iliac artery by iliac/iliac art anastomosis NEC Bypass common iliac artery by aorta/com iliac art anast NEC Bypass iliac artery by aorta/ext iliac art anastomosis NEC Bypass leg artery by aorta/com femoral art anastomosis NEC Bypass leg artery by aorta/deep femoral art anastomosis NEC Bypass iliac artery by iliac/iliac artery anastomosis NEC Emergency bypass of iliac artery by unspecified anastomosis Other specified other bypass of iliac artery Other bypass of iliac artery NOS 7A42.00 Reconstruction of iliac artery 7A42.11 Reconstruction of common iliac artery 7A A A A A42y00 7A42z00 Endarterectomy and patch repair of iliac artery Endarterectomy and patch repair of common iliac artery Endarterectomy of iliac artery NEC Endarterectomy of common iliac artery NEC Other specified reconstruction of iliac artery Reconstruction of iliac artery NOS 7A43.00 Other open operations on iliac artery 7A43.11 Other open operations on common iliac artery 7A A A A A A43y00 7A43z00 Repair of iliac artery NEC Repair of common iliac artery NEC Open embolectomy of iliac artery Open embolectomy of common iliac artery Open insertion of iliac artery stent Other specified other open operation on iliac artery Other open operation on iliac artery NOS 7A44.00 Transluminal operations on iliac artery 7A44.11 Transluminal operations on common iliac artery 7A A A44200 Percutaneous transluminal angioplasty of iliac artery Percutaneous transluminal embolectomy of iliac artery Arteriography of iliac artery 37

74 7A A A A44y00 7A44z00 Arteriography of common iliac artery Insertion of iliac artery stent Percutaneous transluminal insertion of iliac artery stent Other specified transluminal operation on iliac artery Transluminal operation on iliac artery NOS 7A47.00 Other emergency bypass of femoral artery or popliteal artery 7A47.11 Other emerg bypass femoral or popliteal art by anastomosis 7A47.12 Other emergency bypass of common femoral artery 7A47.13 Other emergency bypass of deep femoral artery 7A47.14 Other emergency bypass of popliteal artery 7A47.15 Other emergency bypass of superficial femoral artery 7A47.16 Other emergency bypass of femoral artery 7A A A A A47C00 7A47D00 7A47y00 7A47z00 Emerg bypass popliteal art by pop/pop art anast c prosth NEC Emerg bypass popliteal art by pop/tib art anast c prosth NEC Emerg bypass pop art by pop/tib art anast c vein graft NEC Emerg bypass popliteal art by pop/peron a anast c prosth NEC Emerg bypass femoral artery by fem/fem art anastomosis NEC Emerg bypass popliteal artery by pop/fem art anastomosis NEC Other emergency bypass of femoral or popliteal artery OS Other emergency bypass of femoral or popliteal artery NOS 7A48.00 Other bypass of femoral artery or popliteal artery 7A48.11 Other bypass of femoral or popliteal artery by anastomosis 7A48.12 Other bypass of common femoral artery 7A48.13 Other bypass of deep femoral artery 7A48.14 Other bypass of femoral artery 7A48.15 Other bypass of popliteal artery 7A A A A A A A48600 Bypass femoral artery by fem/pop art anast c prosthesis NEC Bypass popliteal artery by pop/pop a anast c prosthesis NEC Bypass femoral artery by fem/pop art anast c vein graft NEC Bypass popliteal artery by pop/pop a anast c vein graft NEC Bypass femoral artery by fem/tib art anast c prosthesis NEC Bypass popliteal artery by pop/tib a anast c prosthesis NEC Bypass femoral artery by fem/tib art anast c vein graft NEC 38

75 7A A A A48A00 7A48B00 7A48C00 7A48D00 7A48y00 7A48z00 Bypass popliteal artery by pop/tib a anast c vein graft NEC Bypass femoral artery by fem/peron a anast c prosthesis NEC Bypass popliteal artery by pop/peron art anast c prosth NEC Bypass femoral artery by fem/peron a anast c vein graft NEC Bypass popliteal art by pop/peron art anast c vein graft NEC Bypass femoral artery by femoral/femoral art anastomosis NEC Bypass popliteal artery by pop/fem artery anastomosis NEC Other bypass of femoral artery or popliteal artery OS Other bypass of femoral artery or popliteal artery NOS 7A49.00 Reconstruction of femoral artery or popliteal artery 7A49.11 Reconstruction of common femoral artery 7A49.12 Reconstruction of deep femoral artery 7A49.13 Reconstruction of femoral artery 7A49.14 Reconstruction of popliteal artery 7A49.15 Reconstruction of superficial femoral artery 7A A A A A A A A A A A49y00 7A49z00 7A4A.00 7A4A.11 7A4A.12 7A4A.13 7A4A.14 7A4A000 Endarterectomy and patch repair of femoral artery Endarterectomy and patch repair of popliteal artery Endarterectomy of femoral artery NEC Endarterectomy of popliteal artery NEC Profundoplasty femoral artery & patch repair deep fem artery Profundoplasty and patch repair of popliteal artery Profundoplasty of femoral artery NEC Profundoplasty of popliteal artery NEC Reconstruction of femoral artery with vein graft Reconstruction of popliteal artery with vein graft Reconstruction of femoral or popliteal artery OS Reconstruction of femoral or popliteal artery NOS Other open operations on femoral artery or popliteal artery Other open operations on common femoral artery Other open operations on deep femoral artery Other open operations on popliteal artery Other open operations on superficial femoral artery Repair of femoral artery NEC 39

76 7A4A100 7A4A200 7A4A211 7A4A300 7A4A311 7A4A600 7A4A700 7A4A800 7A4Ay00 7A4Az00 7A4B.00 7A4B.11 7A4B.12 7A4B.13 7A4B.14 7A4B.15 7A4B000 7A4B100 7A4B200 7A4B300 7A4B400 7A4B500 7A4B900 7A4By00 7A4Bz00 7A4y.00 7A4z.00 7A A A A53400 Repair of popliteal artery NEC Open embolectomy of femoral artery Open thrombectomy of femoral artery Open embolectomy popliteal artery Open thrombectomy of popliteal artery Operation on popliteal artery NEC Repair of femoral artery with temporary silastic shunt Repair of popliteal artery with temporary silastic shunt Other open operation on femoral or popliteal artery OS Other open operation on femoral or popliteal artery NOS Transluminal operations on femoral or popliteal artery Transluminal procedure on common femoral artery Transluminal procedure on deep femoral artery Transluminal procedure on femoral artery Transluminal procedure on popliteal artery Transluminal procedure on superficial femoral artery Percutaneous transluminal angioplasty of femoral artery Percutaneous transluminal angioplasty of popliteal artery Percutaneous transluminal embolectomy of femoral artery Percutaneous transluminal embolectomy of popliteal artery Percutaneous transluminal embolisation of femoral artery Percutaneous transluminal embolisation of popliteal artery Percutaneous transluminal insertion of stent femoral artery Transluminal operation on femoral or popliteal artery OS Transluminal operation on femoral or popliteal artery NOS Other specified operations on iliac and femoral artery Iliac and femoral artery operations NOS Revision of reconstruction involving iliac artery Revision of reconstruction involving femoral artery Revision of reconstruction of popliteal artery Operation on aneurysm of artery NEC 7A56.00 Other therapeutic transluminal operations on artery 7A56000 Percutaneous transluminal arterial thrombolysis reconstruct 40

77 7A A A A56y00 7A56z00 7M1D300 7M25M00 7N N N N N N N N N N N46B00 7N46C00 7N46D00 7N46F00 7N46G00 7N46H00 7N46J00 7NB8000 7NB8100 7NB8300 7NB8400 7NB8500 7NB8600 Percutaneous transluminal stent reconstruction of artery Percutaneous transluminal balloon angioplasty of artery Percutaneous transluminal placement peripheral stent artery Other specified other therapeutic transluminal operat artery Other therapeutic transluminal operations on artery NOS Transluminal approach to organ through femoral artery Deep circumflex iliac artery flap [SO]Common iliac artery [SO]Internal iliac artery [SO]Common femoral artery [SO]Deep femoral artery [SO]Profunda femoris artery [SO]Superficial femoral artery [SO]Popliteal artery [SO]Tibial artery [SO]External iliac artery [SO]Deep circumflex iliac artery [SO]Other artery of thigh [SO]Anterior tibial artery [SO]Posterior tibial artery [SO]Medial plantar artery [SO]Lateral plantar artery [SO]Dorsalis pedis artery [SO]Digital artery of toe [SO]Anterior tibial artery [SO]Posterior tibial artery [SO]Dorsalis pedis artery [SO]External iliac artery [SO]Iliac artery NEC [SO]Tibial artery NEC F Phantom limb syndrome F F Phantom limb syndrome with pain Phantom limb syndrome without pain 41

78 42

79 Table S3. Comparison of post-propensity match baseline characteristics between the US Truven MarketScan mortality subset and US Truven MarketScan total population US Truven MarketScan Mortality subset US Truven MarketScan Total data Difference Mortality vs Total SGLT-2i (N=71,632) ogld (N=71,632) SGLT-2i (N=116,899) ogld (N=116,899) SGLT-2i ogld n % n % n % n % % % Age in years, mean (SD) 55.8 (9.5) 56.0 (10.2) (9.68) (10.24) Women 31, , , , Frailty (yes) 2, , , , Established CVD* 6, , , , Acute MI , , Unstable angina , , Heart failure 1, , , , Atrial fibrillation 1, , , , Stroke 2, , , , PAD 1, , , , Microvascular disease 18, , , , CKD 1, , , , Statin therapy 48, , , , Antihypertensive therapy 59, , , , Loop diuretics 5, , , , Metformin 59, , , , SU 29, , , , DPP-4 inhibitor 26, , , , TZD 7, , , , GLP-1RA 14, , , , Insulin 21, , , , Index year: , , , , , , , , , , , , Index medication: Amylin analog NA NA NA NA Metformin NA NA 8, NA NA 13, SU NA NA 12, NA NA 20,

80 DPP-4i NA NA 11, NA NA 18, TZD NA NA 4, NA NA 6, GLP-1RA NA NA 9, NA NA 16, Insulin NA NA 23, NA NA 38, Meglitinides NA NA NA NA 1, Acarbose, NA NA NA NA Dapagliflozin 16, NA NA 25, NA NA 0.3 Empagliflozin 5, NA NA 9, NA NA 0.3 Canagliflozin 49, NA NA 81, NA NA -0.5 *Defined as myocardial infarction, unstable angina, stroke, heart failure, transient ischemic attack, coronary revascularization (coronary artery bypass grafting or percutaneous coronary intervention) or occlusive peripheral artery disease on or ever prior to start; CKD=chronic kidney disease; CVD=cardiovascular disease; DPP-4=dipeptidyl peptidase-4; ogld=other glucose-lowering drug; GLP-1RA= Glucagon-like peptide-1 receptor agonists; MI=myocardial infarction; NA=not applicable; PAD=peripheral arterial disease; SGLT-2i=sodium-glucose cotransporter-2 inhibitor; SU=sulfonylurea; TZD=thiazolidinedione; US=United States 44

81 Table S4. List of variables used to develop propensity score Gender Age Race (Caucasian, Black, Asian, Other) Migrant Frailty (yes) Index year Duration of Type 2 diabetes (years) Cardiovascular history Smoker HbA1c >7% (UK only) Body mass index 30 kg/m 2 (UK only) Duration of Type 2 diabetes Estimated Glomerular Filtration Rate <60 ml/min/1.73 m 2 (UK only) Acute myocardial infarction Stroke Heart failure Unstable angina Atrial fibrillation Peripheral artery disease Hypertension Coronary revascularization Coronary artery bypass grafting Percutaneous coronary intervention Carotid intervention Chronic kidney disease Microvascular disease Nephropathy Peripheral neuropathy 45

82 Retinopathy Bariatric surgery Metformin Sulfonylurea DPP-4 inhibitor thiazolidinedione GLP-1 receptor agonist Insulin Statin therapy Antihypertensive therapy P2Y12 inhibitiors Antiplatelets Warfarin Anticoagulants Low dose acetylsalicylic acid Beta-blocker Loop diuretics Thiazides Aldosterone antagonists Weight loss drugs DPP-4=dipeptidyl peptidase-4; GLP-1= Glucagon-like peptide-1. The full list of variables for each individual country is dependent on the variables available in each database 46

83 Table S5. Baseline characteristics for all countries/databases pre-match US Truven MarketScan Norway national registers Denmark national registers Sweden national registers UK CPRD/THIN Germany DPV SGLT-2i N=123,648 ogld N=712,426 SGLT-2i N=14,438 ogld N=96,947 SGLT-2i N=9522 ogld N=119,137 SGLT-2i N=9337 ogld N=200,284 SGLT-2i N=7556 ogld N=73,436 SGLT-2i N=1532 ogld N=23,991 Age, years, mean Women Established CVD* Acute MI Unstable angina HF AF Stroke PAD Microvascular disease CKD Frailty(yes) Glucose-lowering therapies MET SU DPP-4i TZD GLP-1RA Insulin Cardiovascular therapies Antihypertensive

84 therapies Loop diuretics Thiazide diuretics Statin therapy Index year Data are % unless otherwise stated; *Defined as myocardial infarction, unstable angina, stroke, heart failure, transient ischemic attack, coronary revascularization (coronary artery bypass grafting or percutaneous coronary intervention) or occlusive peripheral artery disease on or ever prior to start CKD in the indicated database does not include end stage renal disease; Frailty in the indicated database is defined as 1 hospitalization within 1 year prior to or on index date; Includes angiotensin converting enzyme inhibitors, angiotensin receptor blockers, Ca 2+ channel blockers β-blockers, thiazides; AF=atrial fibrillation; CKD=chronic kidney disease; CPRD, Clinical Practice Research Datalink; CVD=cardiovascular disease; DPP-4i=dipeptidyl peptidase-4 inhibitor; DPV, Diabetes Patientenverlaufsdokumentation; egfr=estimated Glomerular Filtration Rate; HF=heart failure; ogld=other glucose-lowering drug; GLP-1RA= Glucagon-like peptide-1 receptor agonists; MET=metformin; MI, myocardial infarction; PAD=peripheral arterial disease; SGLT-2i=sodium-glucose cotransporter-2 inhibitor; SU=sulfonylurea; THIN, The Health Improvement Network; TZD, thiazolidinedione

85 Table S6. Baseline characteristics for all countries/databases post-match US Truven MarketScan Norway national registers Denmark national registers Sweden national registers UK CPRD/THIN Germany DPV SGLT-2i N=116,899 ogld N=116,899 SGLT-2i N=12,525 ogld N=12,525 SGLT-2i N=9234 ogld N=9234 SGLT-2i N=9189 ogld N=9189 SGLT-2i N=5231 ogld N=5231 SGLT-2i N=1450 ogld N=1450 Age, years, mean (SD) 55.8 (9.7) 55.8 (10.2) 60.4 (11.3) 60.3 (12.7) 60.4 (10.8) 60.9 (12.4) 61.5 (10.2) 61.4 (11.5) 58.8 (10.5) 58.8 (10.8) 61.7 (11.2) 61.7 (11.7) Women Established CVD* Acute MI Unstable angina HF AF Stroke PAD Microvascular disease CKD Frailty(yes) Glucose-lowering therapies MET SU DPP-4i TZD GLP-1RA Insulin Cardiovascular therapies Antihypertensive therapy Loop diuretics Thiazide diuretics

86 ACE inhibitors ARBs Statin therapy Aspirin NA NA Index year Data are % unless otherwise stated; *Defined as myocardial infarction, unstable angina, stroke, heart failure, transient ischemic attack, coronary revascularization (coronary artery bypass grafting or percutaneous coronary intervention) or occlusive peripheral artery disease on or ever prior to start CKD in the indicated database does not include end stage renal disease; Frailty in the indicated database is defined as 1 hospitalization within 1 year prior to or on index date; Includes ACE inhibitors, ARBs, Ca 2+ channel blockers β-blockers, thiazides; ACE=angiotensin converting enzyme; ARB=angiotensin receptor blocker; AF=atrial fibrillation; CKD=chronic kidney disease; CPRD, Clinical Practice Research Datalink; CV=cardiovascular; CVD=cardiovascular disease; DPP-4i=dipeptidyl peptidase-4 inhibitor; DPV, Diabetes Patientenverlaufsdokumentation; egfr=estimated Glomerular Filtration Rate; HF=heart failure; ogld=other glucose-lowering drug; GLP-1RA= Glucagon-like peptide-1 receptor agonists; MET=metformin; MI, myocardial infarction; PAD=peripheral arterial disease; SGLT-2i=sodium-glucose cotransporter-2 inhibitor; SU=sulfonylurea; THIN, The Health Improvement Network; TZD, thiazolidinedione

87 Table S7. Standardized differences in baseline characteristics between SGLT-2i and ogld treatment groups pre- and post-propensity match by country Standardized Difference (%) US Truven MarketScan Norway national registers Denmark national registers Sweden national UK CPRD/THIN Germany DPV (N=233,798) (N=25,050) (N=18,468) registers (N=18,378) (N=10,462) (N=2900) Pre-match Post-match Pre-match Post-match Pre-match Post-match Pre-match Post-match Pre-match Post-match Pre-match Post-match Age Women Established CVD* NA NA Acute MI Unstable angina Heart failure Stroke PAD Coronary revascularization NA NA NA NA NA NA CABG NA NA NA NA PCI NA NA NA NA Carotid intervention NA NA NA NA NA NA 4.3 NA Atrial fibrillation Microvascular disease NA NA

88 CKD NA NA Frailty(yes) Migrant NA NA NA NA NA NA NA NA NA NA Smoker (yes) NA NA NA NA NA NA HbA1c (mean) NA NA NA NA NA NA NA NA HbA1c >7% NA NA NA NA NA NA NA NA BMI (mean) NA NA NA NA NA NA NA NA BMI 30 NA NA NA NA NA NA NA NA Total cholesterol NA NA NA NA NA NA NA NA egfr (mean) NA NA NA NA NA NA NA NA egfr <60 NA NA NA NA NA NA NA NA Nephropathy NA NA NA NA NA NA NA NA Peripheral neuropathy NA NA NA NA NA NA Retinopathy NA NA NA NA NA NA Hypertension NA NA NA NA NA NA NA NA Microalbuminuria NA NA NA NA NA NA NA NA NA NA Weight-loss drugs Bariatric surgery NA NA NA NA NA NA Time from start NA NA NA NA NA NA NA NA T2D duration NA NA Glucose-lowering therapies 52

89 Metformin Sulfonylurea DPP-4 inhibitor TZD NA NA NA NA GLP-1 RA Insulin Cardiovascular therapies Antihypertensive therapy Loop diuretics Thiazide diuretics β-blocker Statin therapy Aspirin NA NA PY12 inhibitors NA NA NA NA Antiplatelets NA NA NA NA NA NA NA NA Warfarin NA NA Anticoagulants NA NA NA NA NA NA NA NA Aldosterone antagonists *Defined as myocardial infarction, unstable angina, stroke, heart failure, transient ischemic attack, coronary revascularization (coronary artery bypass grafting or percutaneous coronary intervention) or occlusive peripheral artery disease on or ever prior to start ; CKD in UK CPRD/THIN and US Truven MarketScan does not include end stage renal disease; Frailty in UK CPRD/THIN is defined as 1 hospitalization within 1 year prior to index date; Includes angiotensin converting enzyme inhibitors, angiotensin receptor blockers, Ca 2+ channel blockers β-blockers, thiazides; BMI=body mass index; CABG=Coronary artery bypass 53

90 grafting; CKD=chronic kidney disease; CPRD, Clinical Practice Research Datalink; CV=cardiovascular; DPP-4=dipeptidyl peptidase-4; DPV=Diabetes Patientenverlaufsdokumentation; egfr=estimated Glomerular Filtration Rate; GLP-1RA= Glucagon-like peptide-1 receptor agonist; MI=myocardial infarction; OAD=oral antidiabetic drugs; PAD=peripheral arterial disease; PCI=percutaneous coronary intervention; T2D=Type 2 diabetes; THIN=The Health Improvement Network; TZD=thiazolidinedione; UK=United Kingdom; US=United States; NA=not applicable (some characteristics were not available in every country) 54

91 Table S8. Composition of SGLT-2 inhibitor class in propensity matched cohorts SGLT-2 inhibitor patients, N Proportions by agent Canagliflozin, % Dapagliflozin, % Empagliflozin, % Hospitalization for heart failure analysis US Truven MarketScan 116, Norway national registers 12, Denmark national registers Sweden national registers UK CPRD/THIN Germany DPV Total 154, All-cause death analysis, and composite (HHF or all-cause death) analysis US Truven MarketScan 71, Norway national registers 12, Denmark national registers Sweden national registers UK CPRD/THIN Total 107,

92 Table S9. Index glucose-lowering medication classes for patients in the other GLD group: all countries combined (HHF analysis) Variable Index medication*, n (%) MET SU DPP-4 inhibitor TZD GLP-1 receptor agonist Insulin Acarbose Amylin analog Meglitinides Other GLD (N=154,528) 17,899 (11.6) 26,203 (17.0) 26,957 (17.4) 7496 (4.9) 21,199 (13.7) 51,838 (33.5) 708 (0.5) 132 (0.1) 2096 (1.4) *In cases of 2 medications, one was randomly selected; DPP-4=dipeptidyl peptidase-4; GLD=glucose-lowering drug; GLP- 1= Glucagon-like peptide-1; HHF=hospitalization for heart failure; MET=metformin; SU=sulfonylurea; TZD= thiazolidinedione 56

93 Table S10. Index glucose-lowering medication classes for patients in the other GLD group: all countries combined (all-cause death analysis and HHF or all-cause death analysis) Other GLD Variable (N=107,811) Index medication*, n (%) MET 12,134 (11.3) SU 18,203 (16.9) DPP-4 inhibitor 19,665 (18.2) TZD 4731 (4.4) GLP-1 receptor agonist 14,743 (13.7) Insulin 36,432 (33.8) Acarbose 424 (0.4) Amylin analog 78 (0.1) Meglitinides 1401 (1.3) *In cases of 2 medications, one was randomly selected; DPP-4=dipeptidyl peptidase-4; GLD=glucose-lowering drug; GLP- 1= Glucagon-like peptide-1; HHF=hospitalization for heart failure; MET=metformin; SU=sulfonylurea; TZD= thiazolidinedione 57

94 Table S11: Number of patients, person-years at risk (on treatment), events and events/100 person-years (incidence rates) for each of the endpoints Country/database Patients, N Person-years Events, n Events / Hospitalization for heart failure 100 personyears US Truven MarketScan Norway national registers Denmark national registers Sweden national registers 233, , ,050 25, ,468 17, ,378 13, UK CPRD/THIN 10, Germany DPV Total 309, , All-cause death US Truven MarketScan Norway national registers Denmark national registers Sweden national registers 143,264 80, ,050 25, ,468 17, ,378 23, UK CPRD/THIN 10, Total 215, , Composite of hospitalization for heart failure or all-cause death US Truven MarketScan Norway national registers Denmark national registers Sweden national registers 143,264 80, ,050 25, ,468 17, ,378 13, UK CPRD/THIN 10, Total 215, , CPRD= Clinical Practice Research Datalink; DPV= Diabetes Patientenverlaufsdokumentation; SGLT-2=sodium-glucose cotransporter-2; THIN=The Health Improvement Network; UK=United Kingdom; US=United States 58

95 Table S12. Pooled event rates by treatment groups SGLT-2i ogld Events Person-years Events/100 person-years Events Person-years Events/100 person-years Hospitalization for heart failure , , All-cause death , , Hospitalization for heart failure and allcause death , ,

96 Table S13. Mean on treatment follow-up time (in days) by country and endpoint HHF ACD HHF/ACD SGLT-2i ogld SGLT-2i ogld SGLT-2i ogld US MarketScan/Truven UK-CPRD/THIN Sweden Norway Denmark Germany Total ACD=all-cause death; HHF=hospitalization for heart failure; ogld=other glucose-lowering drug; SGLT- 2i=sodium-glucose cotransporter-2 60

97 Table S14. Sensitivity analyses examining association between treatment with SGLT-2i vs. ogld and outcomes of HHF in Sweden, among patients with both in- and outpatient hospital visits with primary diagnosis of heart failure, and those with only inpatient hospital visits with the primary diagnosis of heart failure. ogld Events, n (events/100py) SGLT-2i Events, n (events/100py) Hazard ratio 95% CI P-value On treatment, both in- and outpatient 122 (17.1) 69 (10.5) On treatment, only inpatient 70 (9.8) 30 (4.6) <0.001 ITT, both in- and outpatient 150 (17.1) 101 (12.3) ITT, only inpatient 89 (10.1) 49 (5.9) CI=confidence interval; HHF=hospitalization for heart failure; ogld=other glucose-lowering drug; PY=personyears; SGLT-2i=sodium-glucose cotransporter-2. 61

98 SUPPLEMENTARY FIGURES Figure S1: Patient selection flow-charts for each country A. US Truven MarketScan B. Norway National Registers 62

99 C. Denmark National Registers D. Sweden National Registers 63

100 E. UK CPRD/THIN F. Germany DPV CPRD=Clinical Practice Research Datalink; DPV= Diabetes Patientenverlaufsdokumentation; GLD=glucose-lowering drugs; HHF=hospitalization for heart failure; SGLT-2i=sodium-glucose cotransporter-2 inhibitors; THIN=The Health Improvement Network; UK=United Kingdom; US=United States 64

101 Figure S2: Propensity score distribution by country pre-match A B C D E F CPRD=Clinical Practice Research Datalink; DPV=Diabetes Patientenverlaufsdokumentation; SGLT-2=sodium-glucose cotransporter-2; THIN=The Health Improvement Network; UK=United Kingdom; US=United States 65

102 Figure S3: Propensity score distribution by country post-match A B C D E F CPRD=Clinical Practice Research Datalink; DPV=Diabetes Patientenverlaufsdokumentation; SGLT-2=sodium-glucose cotransporter-2; THIN=The Health Improvement Network; UK=United Kingdom; US=United States 66

103 Figure S4: Standardized difference between SGLT-2 inhibitor and other GLD treatment groups by country A. US Truven MarketScan 67

104 B. Norway National Registers 68

105 C. Denmark National Registers 69

106 D. Sweden National Registers 70

107 E. UK CPRD/THIN 71

108 72

109 F. Germany DPV DPP-4=dipeptidyl peptidase-4; egfr=estimated Glomerular Filtration Rate; GLP-1= Glucagon-like peptide-1; ogld=other glucose-lowering drug; SGLT-2i=sodium-glucose cotransporter -2 inhibitor; T2D=Type 2 diabetes 73

110 Figure S5: Stepwise sensitivity analysis (sequentially removing comparators): Outcome of hospitalization for heart failure A. On treatment, adjusted* TZD removed *Adjusted for history of heart failure, age, gender, frailty, history of myocardial infarction, history of atrial fibrillation, hypertension, obesity / body mass index, duration of diabetes, ACE inhibitor or ARB use; β-blocker or α-blocker use, Ca + -channel blocker use, loop diuretic use, thiazide diuretic use; TZD=thiazolidinedione 74

111 B. On treatment, adjusted* TZD and insulin removed *Adjusted for history of heart failure, age, gender, frailty, history of myocardial infarction, history of atrial fibrillation, hypertension, obesity / body mass index, duration of diabetes, ACE inhibitor or ARB use; β-blocker or α-blocker use, Ca + -channel blocker use, loop diuretic use, thiazide diuretic use; TZD= thiazolidinedione 75

112 C. On treatment, adjusted* TZD, insulin and SU removed *Adjusted for history of heart failure, age, gender, frailty, history of myocardial infarction, history of atrial fibrillation, hypertension, obesity / body mass index, duration of diabetes, ACE inhibitor or ARB use; β-blocker or α-blocker use, Ca + -channel blocker use, loop diuretic use, thiazide diuretic use; SU=sulfonylurea; TZD= thiazolidinedione 76

113 Figure S6: Outcome of hospitalization for heart failure excluding patients with baseline GLP-1 RAs, on treatment, adjusted* *Adjusted for history of heart failure, age, gender, frailty, history of myocardial infarction, history of atrial fibrillation, hypertension, obesity / body mass index, duration of diabetes, ACE inhibitor or ARB use; β-blocker or α-blocker use, Ca + -channel blocker use, loop diuretic use, thiazide diuretic use; SU=sulfonylurea; TZD= thiazolidinedione Data from Germany and UK not included as they contributed too few HHF events after removal of GLP1-RA class 77

114 Figure S7: Outcomes for hospitalization for heart failure, all-cause death, and hospitalization for heart failure or all-cause death for the SGLT-2 inhibitor versus other GLD treatment groups A. US only B. European countries combined P-value for SGLT-2i versus ogld comparison <0.001 for all endpoints; ogld=other glucose-lowering drug; 78